Keyword: machine learning
2024 (1)
Advanced Computer Science Applications: Recent Trends in AI, Machine Learning, and Network Security. Singh, K., Banda, L., & Manjul, M. Apple Academic Press Inc., Boca Raton, FL, 2024.
abstract bibtex
abstract bibtex
Discusses design algorithms that allow computers to employ machine learning to display behavior learned from past experiences for solutions to security issues in data management. Algorithms discussed include convolutional neural network, random forest algorithm, k-nearest neighbor algorithm, Apriori algorithm, MapReduce algorithm, etc.
@book{singh_advanced_2024,
address = {Boca Raton, FL},
title = {Advanced {Computer} {Science} {Applications}: {Recent} {Trends} in {AI}, {Machine} {Learning}, and {Network} {Security}},
isbn = {978-1-00-336906-6},
shorttitle = {Advanced {Computer} {Science} {Applications}},
abstract = {Discusses design algorithms that allow computers to employ machine learning to display behavior learned from past experiences for solutions to security issues in data management. Algorithms discussed include convolutional neural network, random forest algorithm, k-nearest neighbor algorithm, Apriori algorithm, MapReduce algorithm, etc.},
language = {eng},
publisher = {Apple Academic Press Inc.},
author = {Singh, Karan and Banda, Latha and Manjul, Manisha},
year = {2024},
keywords = {Artificial intelligence, Computer networks, Machine learning, Security measures},
}2023 (10)
Is It Worth It? Comparing Six Deep and Classical Methods for Unsupervised Anomaly Detection in Time Series. Rewicki, F., Denzler, J., & Niebling, J. Applied Sciences, 13(3):1778, January, 2023. Number: 3 Publisher: Multidisciplinary Digital Publishing Institute
![Is It Worth It? Comparing Six Deep and Classical Methods for Unsupervised Anomaly Detection in Time Series [link]](https://bibbase.org/img/filetypes/link.svg "link")
Paper doi abstract bibtex
Paper doi abstract bibtex Detecting anomalies in time series data is important in a variety of fields, including system monitoring, healthcare and cybersecurity. While the abundance of available methods makes it difficult to choose the most appropriate method for a given application, each method has its strengths in detecting certain types of anomalies. In this study, we compare six unsupervised anomaly detection methods of varying complexity to determine whether more complex methods generally perform better and if certain methods are better suited to certain types of anomalies. We evaluated the methods using the UCR anomaly archive, a recent benchmark dataset for anomaly detection. We analyzed the results on a dataset and anomaly-type level after adjusting the necessary hyperparameters for each method. Additionally, we assessed the ability of each method to incorporate prior knowledge about anomalies and examined the differences between point-wise and sequence-wise features. Our experiments show that classical machine learning methods generally outperform deep learning methods across a range of anomaly types.
@article{rewicki_is_2023,
title = {Is {It} {Worth} {It}? {Comparing} {Six} {Deep} and {Classical} {Methods} for {Unsupervised} {Anomaly} {Detection} in {Time} {Series}},
volume = {13},
copyright = {http://creativecommons.org/licenses/by/3.0/},
issn = {2076-3417},
shorttitle = {Is {It} {Worth} {It}?},
url = {https://www.mdpi.com/2076-3417/13/3/1778},
doi = {10.3390/app13031778},
abstract = {Detecting anomalies in time series data is important in a variety of fields, including system monitoring, healthcare and cybersecurity. While the abundance of available methods makes it difficult to choose the most appropriate method for a given application, each method has its strengths in detecting certain types of anomalies. In this study, we compare six unsupervised anomaly detection methods of varying complexity to determine whether more complex methods generally perform better and if certain methods are better suited to certain types of anomalies. We evaluated the methods using the UCR anomaly archive, a recent benchmark dataset for anomaly detection. We analyzed the results on a dataset and anomaly-type level after adjusting the necessary hyperparameters for each method. Additionally, we assessed the ability of each method to incorporate prior knowledge about anomalies and examined the differences between point-wise and sequence-wise features. Our experiments show that classical machine learning methods generally outperform deep learning methods across a range of anomaly types.},
language = {en},
number = {3},
urldate = {2023-11-14},
journal = {Applied Sciences},
author = {Rewicki, Ferdinand and Denzler, Joachim and Niebling, Julia},
month = jan,
year = {2023},
note = {Number: 3
Publisher: Multidisciplinary Digital Publishing Institute},
keywords = {Anomaly detection, Time series, Deep Learning, Benchmark, Machine learning},
pages = {1778},
file = {Full Text PDF:C\:\\Users\\Guillaume\\Zotero\\storage\\IU3IXMZ2\\Rewicki et al. - 2023 - Is It Worth It Comparing Six Deep and Classical M.pdf:application/pdf},
}Knowledge Discovery: Methods from data mining and machine learning. Shu, X. & Ye, Y. Social Science Research, 110:102817, 2023.
Paper doi abstract bibtex
Paper doi abstract bibtex The interdisciplinary field of knowledge discovery and data mining emerged from a necessity of big data requiring new analytical methods beyond the traditional statistical approaches to discover new knowledge from the data mine. This emergent approach is a dialectic research process that is both deductive and inductive. The data mining approach automatically or semi-automatically considers a larger number of joint, interactive, and independent predictors to address causal heterogeneity and improve prediction. Instead of challenging the conventional model-building approach, it plays an important complementary role in improving model goodness of fit, revealing valid and significant hidden patterns in data, identifying nonlinear and non-additive effects, providing insights into data developments, methods, and theory, and enriching scientific discovery. Machine learning builds models and algorithms by learning and improving from data when the explicit model structure is unclear and algorithms with good performance are difficult to attain. The most recent development is to incorporate this new paradigm of predictive modeling with the classical approach of parameter estimation regressions to produce improved models that combine explanation and prediction.
@article{SHU2023102817,
title = {Knowledge Discovery: Methods from data mining and machine learning},
journal = {Social Science Research},
volume = {110},
pages = {102817},
year = {2023},
issn = {0049-089X},
doi = {https://doi.org/10.1016/j.ssresearch.2022.102817},
url = {https://www.sciencedirect.com/science/article/pii/S0049089X22001284},
author = {Xiaoling Shu and Yiwan Ye},
keywords = {Knowledge discovery, data mining, machine learning, causal discovery, predition, big data},
abstract = {The interdisciplinary field of knowledge discovery and data mining emerged from a necessity of big data requiring new analytical methods beyond the traditional statistical approaches to discover new knowledge from the data mine. This emergent approach is a dialectic research process that is both deductive and inductive. The data mining approach automatically or semi-automatically considers a larger number of joint, interactive, and independent predictors to address causal heterogeneity and improve prediction. Instead of challenging the conventional model-building approach, it plays an important complementary role in improving model goodness of fit, revealing valid and significant hidden patterns in data, identifying nonlinear and non-additive effects, providing insights into data developments, methods, and theory, and enriching scientific discovery. Machine learning builds models and algorithms by learning and improving from data when the explicit model structure is unclear and algorithms with good performance are difficult to attain. The most recent development is to incorporate this new paradigm of predictive modeling with the classical approach of parameter estimation regressions to produce improved models that combine explanation and prediction.}
}Winning the CityLearn Challenge: Adaptive Optimization with Evolutionary Search under Trajectory-based Guidance. Khattar, V. & Jin, M. In AAAI Conference on Artificial Intelligence (AAAI) AI for Social Impact Track, 2023.
Arxiv ![Winning the CityLearn Challenge: Adaptive Optimization with Evolutionary Search under Trajectory-based Guidance [pdf]](https://bibbase.org/img/filetypes/pdf.svg "pdf")
Pdf abstract bibtex 25 downloads
Arxiv Pdf abstract bibtex 25 downloads Modern power systems will have to face difficult challenges in the years to come: frequent blackouts in urban areas caused by high peaks of electricity demand, grid instability exacerbated by the intermittency of renewable generation, and climate change on a global scale amplified by increasing carbon emissions. While current practices are growingly inadequate, the pathway of artificial intelligence (AI)-based methods to widespread adoption is hindered by missing aspects of trustworthiness. The CityLearn Challenge is an exemplary opportunity for researchers from multi-disciplinary fields to investigate the potential of AI to tackle these pressing issues within the energy domain, collectively modeled as a reinforcement learning (RL) task. Multiple real-world challenges faced by contemporary RL techniques are embodied in the problem formulation. In this paper, we present a novel method using the solution function of optimization as policies to compute the actions for sequential decision-making, while notably adapting the parameters of the optimization model from online observations. Algorithmically, this is achieved by an evolutionary algorithm under a novel trajectory-based guidance scheme. Formally, the global convergence property is established. Our agent ranked first in the latest 2021 CityLearn Challenge, being able to achieve superior performance in almost all metrics while maintaining some key aspects of interpretability.
@inproceedings{2023_4C_CL,
title={Winning the CityLearn Challenge: Adaptive Optimization with Evolutionary Search under Trajectory-based Guidance},
author={Vanshaj Khattar and Ming Jin},
booktitle={AAAI Conference on Artificial Intelligence (AAAI) AI for Social Impact Track},
pages={},
year={2023},
url_arXiv={https://arxiv.org/abs/2212.01939},
url_pdf={ESGuidance_ZOiRL.pdf},
keywords = {Optimization, Power system, Reinforcement learning, Machine Learning},
abstract={Modern power systems will have to face difficult challenges in the years to come: frequent blackouts in urban areas caused by high peaks of electricity demand, grid instability exacerbated by the intermittency of renewable generation, and climate change on a global scale amplified by increasing carbon emissions. While current practices are growingly inadequate, the pathway of artificial intelligence (AI)-based methods to widespread adoption is hindered by missing aspects of trustworthiness. The CityLearn Challenge is an exemplary opportunity for researchers from multi-disciplinary fields to investigate the potential of AI to tackle these pressing issues within the energy domain, collectively modeled as a reinforcement learning (RL) task. Multiple real-world challenges faced by contemporary RL techniques are embodied in the problem formulation. In this paper, we present a novel method using the solution function of optimization as policies to compute the actions for sequential decision-making, while notably adapting the parameters of the optimization model from online observations. Algorithmically, this is achieved by an evolutionary algorithm under a novel trajectory-based guidance scheme. Formally, the global convergence property is established. Our agent ranked first in the latest 2021 CityLearn Challenge, being able to achieve superior performance in almost all metrics while maintaining some key aspects of interpretability. },
}ChatCAD: Interactive Computer-Aided Diagnosis on Medical Image using Large Language Models. Wang, S., Zhao, Z., Ouyang, X., Wang, Q., & Shen, D. arXiv.org, February, 2023. Place: Ithaca Publisher: Cornell University Library, arXiv.org
Paper abstract bibtex
Paper abstract bibtex Large language models (LLMs) have recently demonstrated their potential in clinical applications, providing valuable medical knowledge and advice. For example, a large dialog LLM like ChatGPT has successfully passed part of the US medical licensing exam. However, LLMs currently have difficulty processing images, making it challenging to interpret information from medical images, which are rich in information that supports clinical decisions. On the other hand, computer-aided diagnosis (CAD) networks for medical images have seen significant success in the medical field by using advanced deep-learning algorithms to support clinical decision-making. This paper presents a method for integrating LLMs into medical-image CAD networks. The proposed framework uses LLMs to enhance the output of multiple CAD networks, such as diagnosis networks, lesion segmentation networks, and report generation networks, by summarizing and reorganizing the information presented in natural language text format. The goal is to merge the strengths of LLMs' medical domain knowledge and logical reasoning with the vision understanding capability of existing medical-image CAD models to create a more user-friendly and understandable system for patients compared to conventional CAD systems. In the future, LLM's medical knowledge can be also used to improve the performance of vision-based medical-image CAD models.
@article{wang_chatcad_2023,
title = {{ChatCAD}: {Interactive} {Computer}-{Aided} {Diagnosis} on {Medical} {Image} using {Large} {Language} {Models}},
url = {https://www.proquest.com/working-papers/chatcad-interactive-computer-aided-diagnosis-on/docview/2776851931/se-2},
abstract = {Large language models (LLMs) have recently demonstrated their potential in clinical applications, providing valuable medical knowledge and advice. For example, a large dialog LLM like ChatGPT has successfully passed part of the US medical licensing exam. However, LLMs currently have difficulty processing images, making it challenging to interpret information from medical images, which are rich in information that supports clinical decisions. On the other hand, computer-aided diagnosis (CAD) networks for medical images have seen significant success in the medical field by using advanced deep-learning algorithms to support clinical decision-making. This paper presents a method for integrating LLMs into medical-image CAD networks. The proposed framework uses LLMs to enhance the output of multiple CAD networks, such as diagnosis networks, lesion segmentation networks, and report generation networks, by summarizing and reorganizing the information presented in natural language text format. The goal is to merge the strengths of LLMs' medical domain knowledge and logical reasoning with the vision understanding capability of existing medical-image CAD models to create a more user-friendly and understandable system for patients compared to conventional CAD systems. In the future, LLM's medical knowledge can be also used to improve the performance of vision-based medical-image CAD models.},
language = {English},
journal = {arXiv.org},
author = {Wang, Sheng and Zhao, Zihao and Ouyang, Xi and Wang, Qian and Shen, Dinggang},
month = feb,
year = {2023},
note = {Place: Ithaca
Publisher: Cornell University Library, arXiv.org},
keywords = {Machine learning, Business And Economics--Banking And Finance, Computer Vision and Pattern Recognition, Natural language processing, Algorithms, Image enhancement, Diagnosis, Cognition \& reasoning, Decision making, Computer aided decision processes, Image and Video Processing, Image segmentation, Medical imaging, Networks},
}FMG-Net and W-Net: Multigrid Inspired Deep Learning Architectures For Medical Imaging Segmentation. Celaya, A., Riviere, B., & Fuentes, D. Neurips 2023, 2023.
bibtex
bibtex
@article{Celaya2023,
title={FMG-Net and W-Net: Multigrid Inspired Deep Learning Architectures For Medical Imaging Segmentation},
author={Celaya, A. and Riviere, B. and Fuentes, D.},
journal={Neurips 2023},
year={2023},
Keywords={Machine Learning, Image Segmentation, Mathematical Biology, Data Science}
}Distinguishing Human-Written and ChatGPT-Generated Text Using Machine Learning. Alamleh, H., Alqahtani, A., & Elsaid, A. In pages 154–158, 2023.
Paper doi abstract bibtex
Paper doi abstract bibtex The use of sophisticated Artificial Intelligence (AI) language models, including ChatGPT, has led to growing concerns regarding the ability to distinguish between human-written and AI-generated text in academic and scholarly settings. This study seeks to evaluate the effectiveness of machine learning algorithms in differentiating between human-written and AI-generated text. To accomplish this, we collected responses from Computer Science students for both essay and programming assignments. We then trained and evaluated several machine learning models, including Logistic Regression (LR), Decision Trees (DT), Support Vector Machines (SVM), Neural Networks (NN), and Random Forests (RF), based on accuracy, computational efficiency, and confusion matrices. By comparing the performance of these models, we identified the most suitable one for the task at hand. The use of machine learning algorithms for detecting text generated by AI has significant potential for applications in content moderation, plagiarism detection, and quality control for text generation systems, thereby contributing to the preservation of academic integrity in the face of rapidly advancing AI-driven content generation. © 2023 IEEE.
@inproceedings{alamleh_distinguishing_2023,
title = {Distinguishing {Human}-{Written} and {ChatGPT}-{Generated} {Text} {Using} {Machine} {Learning}},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85161905543&doi=10.1109%2fSIEDS58326.2023.10137767&partnerID=40&md5=cf2c79bb7679470ba3e43823f468e077},
doi = {10.1109/SIEDS58326.2023.10137767},
abstract = {The use of sophisticated Artificial Intelligence (AI) language models, including ChatGPT, has led to growing concerns regarding the ability to distinguish between human-written and AI-generated text in academic and scholarly settings. This study seeks to evaluate the effectiveness of machine learning algorithms in differentiating between human-written and AI-generated text. To accomplish this, we collected responses from Computer Science students for both essay and programming assignments. We then trained and evaluated several machine learning models, including Logistic Regression (LR), Decision Trees (DT), Support Vector Machines (SVM), Neural Networks (NN), and Random Forests (RF), based on accuracy, computational efficiency, and confusion matrices. By comparing the performance of these models, we identified the most suitable one for the task at hand. The use of machine learning algorithms for detecting text generated by AI has significant potential for applications in content moderation, plagiarism detection, and quality control for text generation systems, thereby contributing to the preservation of academic integrity in the face of rapidly advancing AI-driven content generation. © 2023 IEEE.},
author = {Alamleh, H. and Alqahtani, A.A.S. and Elsaid, A.},
year = {2023},
keywords = {AI, AI-generated text, ChatGPT, NLP, TF-IDF, TextOriginClassifier, academic integrity, content detection, human-written text, machine learning},
pages = {154--158},
}Cloud-Free Global Maps of Essential Vegetation Traits Processed from the TOA Sentinel-3 Catalogue in Google Earth Engine. Kovács, D. D., Reyes-Muñoz, P., Salinero-Delgado, M., Mészáros, V. I., Berger, K., & Verrelst, J. Remote Sensing, 15(13):3404, January, 2023. Number: 13 Publisher: Multidisciplinary Digital Publishing Institute
Paper doi abstract bibtex
Paper doi abstract bibtex Global mapping of essential vegetation traits (EVTs) through data acquired by Earth-observing satellites provides a spatially explicit way to analyze the current vegetation states and dynamics of our planet. Although significant efforts have been made, there is still a lack of global and consistently derived multi-temporal trait maps that are cloud-free. Here we present the processing chain for the spatiotemporally continuous production of four EVTs at a global scale: (1) fraction of absorbed photosynthetically active radiation (FAPAR), (2) leaf area index (LAI), (3) fractional vegetation cover (FVC), and (4) leaf chlorophyll content (LCC). The proposed workflow presents a scalable processing approach to the global cloud-free mapping of the EVTs. Hybrid retrieval models, named S3-TOA-GPR-1.0-WS, were implemented into Google Earth Engine (GEE) using Sentinel-3 Ocean and Land Color Instrument (OLCI) Level-1B for the mapping of the four EVTs along with associated uncertainty estimates. We used the Whittaker smoother (WS) for the temporal reconstruction of the four EVTs, which led to continuous data streams, here applied to the year 2019. Cloud-free maps were produced at 5 km spatial resolution at 10-day time intervals. The consistency and plausibility of the EVT estimates for the resulting annual profiles were evaluated by per-pixel intra-annually correlating against corresponding vegetation products of both MODIS and Copernicus Global Land Service (CGLS). The most consistent results were obtained for LAI, which showed intra-annual correlations with an average Pearson correlation coefficient (R) of 0.57 against the CGLS LAI product. Globally, the EVT products showed consistent results, specifically obtaining higher correlation than R\textgreater 0.5 with reference products between 30 and 60° latitude in the Northern Hemisphere. Additionally, intra-annual goodness-of-fit statistics were also calculated locally against reference products over four distinct vegetated land covers. As a general trend, vegetated land covers with pronounced phenological dynamics led to high correlations between the different products. However, sparsely vegetated fields as well as areas near the equator linked to smaller seasonality led to lower correlations. We conclude that the global gap-free mapping of the four EVTs was overall consistent. Thanks to GEE, the entire OLCI L1B catalogue can be processed efficiently into the EVT products on a global scale and made cloud-free with the WS temporal reconstruction method. Additionally, GEE facilitates the workflow to be operationally applicable and easily accessible to the broader community.
@article{kovacs_cloud-free_2023,
title = {Cloud-{Free} {Global} {Maps} of {Essential} {Vegetation} {Traits} {Processed} from the {TOA} {Sentinel}-3 {Catalogue} in {Google} {Earth} {Engine}},
volume = {15},
copyright = {http://creativecommons.org/licenses/by/3.0/},
issn = {2072-4292},
url = {https://www.mdpi.com/2072-4292/15/13/3404},
doi = {10.3390/rs15133404},
abstract = {Global mapping of essential vegetation traits (EVTs) through data acquired by Earth-observing satellites provides a spatially explicit way to analyze the current vegetation states and dynamics of our planet. Although significant efforts have been made, there is still a lack of global and consistently derived multi-temporal trait maps that are cloud-free. Here we present the processing chain for the spatiotemporally continuous production of four EVTs at a global scale: (1) fraction of absorbed photosynthetically active radiation (FAPAR), (2) leaf area index (LAI), (3) fractional vegetation cover (FVC), and (4) leaf chlorophyll content (LCC). The proposed workflow presents a scalable processing approach to the global cloud-free mapping of the EVTs. Hybrid retrieval models, named S3-TOA-GPR-1.0-WS, were implemented into Google Earth Engine (GEE) using Sentinel-3 Ocean and Land Color Instrument (OLCI) Level-1B for the mapping of the four EVTs along with associated uncertainty estimates. We used the Whittaker smoother (WS) for the temporal reconstruction of the four EVTs, which led to continuous data streams, here applied to the year 2019. Cloud-free maps were produced at 5 km spatial resolution at 10-day time intervals. The consistency and plausibility of the EVT estimates for the resulting annual profiles were evaluated by per-pixel intra-annually correlating against corresponding vegetation products of both MODIS and Copernicus Global Land Service (CGLS). The most consistent results were obtained for LAI, which showed intra-annual correlations with an average Pearson correlation coefficient (R) of 0.57 against the CGLS LAI product. Globally, the EVT products showed consistent results, specifically obtaining higher correlation than R{\textgreater} 0.5 with reference products between 30 and 60° latitude in the Northern Hemisphere. Additionally, intra-annual goodness-of-fit statistics were also calculated locally against reference products over four distinct vegetated land covers. As a general trend, vegetated land covers with pronounced phenological dynamics led to high correlations between the different products. However, sparsely vegetated fields as well as areas near the equator linked to smaller seasonality led to lower correlations. We conclude that the global gap-free mapping of the four EVTs was overall consistent. Thanks to GEE, the entire OLCI L1B catalogue can be processed efficiently into the EVT products on a global scale and made cloud-free with the WS temporal reconstruction method. Additionally, GEE facilitates the workflow to be operationally applicable and easily accessible to the broader community.},
language = {en},
number = {13},
urldate = {2023-10-16},
journal = {Remote Sensing},
author = {Kovács, Dávid D. and Reyes-Muñoz, Pablo and Salinero-Delgado, Matías and Mészáros, Viktor Ixion and Berger, Katja and Verrelst, Jochem},
month = jan,
year = {2023},
note = {Number: 13
Publisher: Multidisciplinary Digital Publishing Institute},
keywords = {FAPAR, FVC, Gaussian process regression, Google Earth Engine, LAI, LCC, Sentinel-3, TOA radiance, Whittaker, essential vegetation traits, machine learning, temporal reconstruction, time series},
pages = {3404},
}A Human-on-the-Loop Optimization Autoformalism Approach for Sustainability. Jin, M., Sel, B., Hardeep, F., & Yin, W. In Preprint, 2023.
Arxiv bibtex 13 downloads
Arxiv bibtex 13 downloads @inproceedings{2023_5C_Chatenergy,
title={A Human-on-the-Loop Optimization Autoformalism Approach
for Sustainability},
author={Ming Jin and Bilgehan Sel and Fnu Hardeep and Wotao Yin},
booktitle={Preprint},
pages={},
year={2023},
url_arXiv = {https://arxiv.org/abs/2210.06516},
keywords = {Machine Learning, Energy}
}From ChatGPT to CatGPT: The Implications of Artificial Intelligence on Library Cataloging. Brzustowicz, R. Information Technology and Libraries, September, 2023. Number: 3
Paper doi abstract bibtex
Paper doi abstract bibtex This paper explores the potential of language models such as ChatGPT to transform library cataloging. Through experiments with ChatGPT, the author demonstrates its ability to generate accurate MARC records using RDA and other standards such as the Dublin Core Metadata Element Set. These results demonstrate the potential of ChatGPT as a tool for streamlining the record creation process and improving efficiency in library settings. The use of AI-generated records, however, also raises important questions related to intellectual property rights and bias. The paper reviews recent studies on AI in libraries and concludes that further research and development of this innovative technology is necessary to ensure its responsible implementation in the field of library cataloging.
@article{brzustowicz_chatgpt_2023,
title = {From {ChatGPT} to {CatGPT}: {The} {Implications} of {Artificial} {Intelligence} on {Library} {Cataloging}},
volume = {42},
copyright = {Copyright (c) 2023 Richard Brzustowicz},
issn = {2163-5226},
shorttitle = {From {ChatGPT} to {CatGPT}},
url = {https://ital.corejournals.org/index.php/ital/article/view/16295},
doi = {10.5860/ital.v42i3.16295},
abstract = {This paper explores the potential of language models such as ChatGPT to transform library cataloging. Through experiments with ChatGPT, the author demonstrates its ability to generate accurate MARC records using RDA and other standards such as the Dublin Core Metadata Element Set. These results demonstrate the potential of ChatGPT as a tool for streamlining the record creation process and improving efficiency in library settings. The use of AI-generated records, however, also raises important questions related to intellectual property rights and bias. The paper reviews recent studies on AI in libraries and concludes that further research and development of this innovative technology is necessary to ensure its responsible implementation in the field of library cataloging.},
language = {en},
number = {3},
urldate = {2023-11-11},
journal = {Information Technology and Libraries},
author = {Brzustowicz, Richard},
month = sep,
year = {2023},
note = {Number: 3},
keywords = {Machine learning, Artificial intelligence, Automation, Cataloging, ChatGPT, Dublin Core, Information organization, Library cataloging, Library technology, MARC records, Metadata, Natural language processing, RDA},
file = {Full Text PDF:/Users/manika/Zotero/storage/8EGAJENM/Brzustowicz - 2023 - From ChatGPT to CatGPT The Implications of Artifi.pdf:application/pdf},
}GEORGIA: A Graph Neural Network Based EmulatOR for Glacial Isostatic Adjustment. Lin, Y., Whitehouse, P. L., Valentine, A. P., & Woodroffe, S. A. Geophysical Research Letters, 50(18):e2023GL103672, 2023. _eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1029/2023GL103672
Paper doi abstract bibtex
Paper doi abstract bibtex Glacial isostatic adjustment (GIA) modeling is not only useful for understanding past relative sea-level change but also for projecting future sea-level change due to ongoing land deformation. However, GIA model predictions are subject to a range of uncertainties, most notably due to uncertainty in the input ice history. An effective way to reduce this uncertainty is to perform data-model comparisons over a large ensemble of possible ice histories, but this is often impossible due to computational limitations. Here we address this problem by building a deep-learning-based GIA emulator that can mimic the behavior of a physics-based GIA model while being computationally cheap to evaluate. Assuming a single 1-D Earth rheology, our emulator shows 0.54 m mean absolute error on 150 out-of-sample testing data with \textless0.5 s emulation time. Using this emulator, two illustrative applications related to the calculation of barystatic sea level are provided for use by the sea-level community.
@article{lin_georgia_2023,
title = {{GEORGIA}: {A} {Graph} {Neural} {Network} {Based} {EmulatOR} for {Glacial} {Isostatic} {Adjustment}},
volume = {50},
copyright = {© 2023. The Authors. Geophysical Research Letters published by Wiley Periodicals LLC on behalf of American Geophysical Union.},
issn = {1944-8007},
shorttitle = {{GEORGIA}},
url = {https://onlinelibrary.wiley.com/doi/abs/10.1029/2023GL103672},
doi = {10.1029/2023GL103672},
abstract = {Glacial isostatic adjustment (GIA) modeling is not only useful for understanding past relative sea-level change but also for projecting future sea-level change due to ongoing land deformation. However, GIA model predictions are subject to a range of uncertainties, most notably due to uncertainty in the input ice history. An effective way to reduce this uncertainty is to perform data-model comparisons over a large ensemble of possible ice histories, but this is often impossible due to computational limitations. Here we address this problem by building a deep-learning-based GIA emulator that can mimic the behavior of a physics-based GIA model while being computationally cheap to evaluate. Assuming a single 1-D Earth rheology, our emulator shows 0.54 m mean absolute error on 150 out-of-sample testing data with {\textless}0.5 s emulation time. Using this emulator, two illustrative applications related to the calculation of barystatic sea level are provided for use by the sea-level community.},
language = {en},
number = {18},
urldate = {2024-01-29},
journal = {Geophysical Research Letters},
author = {Lin, Yucheng and Whitehouse, Pippa L. and Valentine, Andrew P. and Woodroffe, Sarah A.},
year = {2023},
note = {\_eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1029/2023GL103672},
keywords = {glacial isostatic adjustment, machine learning, sea-level change, statistical emulator},
pages = {e2023GL103672},
}2022 (11)
A Survey of Open Source Automation Tools for Data Science Predictions. Hoell, N. August, 2022. arXiv:2208.11792 [cs]
Paper doi abstract bibtex
Paper doi abstract bibtex We present an expository overview of technical and cultural challenges to the development and adoption of automation at various stages in the data science prediction lifecycle, restricting focus to supervised learning with structured datasets. In addition, we review popular open source Python tools implementing common solution patterns for the automation challenges and highlight gaps where we feel progress still demands to be made.
@misc{hoell_survey_2022,
title = {A {Survey} of {Open} {Source} {Automation} {Tools} for {Data} {Science} {Predictions}},
url = {http://arxiv.org/abs/2208.11792},
doi = {10.48550/arXiv.2208.11792},
abstract = {We present an expository overview of technical and cultural challenges to the development and adoption of automation at various stages in the data science prediction lifecycle, restricting focus to supervised learning with structured datasets. In addition, we review popular open source Python tools implementing common solution patterns for the automation challenges and highlight gaps where we feel progress still demands to be made.},
urldate = {2022-08-29},
publisher = {arXiv},
author = {Hoell, Nicholas},
month = aug,
year = {2022},
note = {arXiv:2208.11792 [cs]},
keywords = {machine learning, mentions sympy, review},
}A Machine Learning Tutorial for Operational Meteorology. Part I: Traditional Machine Learning. Chase, R. J., Harrison, D. R., Burke, A., Lackmann, G. M., & McGovern, A. Weather and Forecasting, 37(8):1509–1529, August, 2022.
Paper doi abstract bibtex
Paper doi abstract bibtex Abstract Recently, the use of machine learning in meteorology has increased greatly. While many machine learning methods are not new, university classes on machine learning are largely unavailable to meteorology students and are not required to become a meteorologist. The lack of formal instruction has contributed to perception that machine learning methods are “black boxes” and thus end-users are hesitant to apply the machine learning methods in their everyday workflow. To reduce the opaqueness of machine learning methods and lower hesitancy toward machine learning in meteorology, this paper provides a survey of some of the most common machine learning methods. A familiar meteorological example is used to contextualize the machine learning methods while also discussing machine learning topics using plain language. The following machine learning methods are demonstrated: linear regression, logistic regression, decision trees, random forest, gradient boosted decision trees, naïve Bayes, and support vector machines. Beyond discussing the different methods, the paper also contains discussions on the general machine learning process as well as best practices to enable readers to apply machine learning to their own datasets. Furthermore, all code (in the form of Jupyter notebooks and Google Colaboratory notebooks) used to make the examples in the paper is provided in an effort to catalyze the use of machine learning in meteorology.
@article{chase_machine_2022,
title = {A {Machine} {Learning} {Tutorial} for {Operational} {Meteorology}. {Part} {I}: {Traditional} {Machine} {Learning}},
volume = {37},
issn = {0882-8156, 1520-0434},
shorttitle = {A {Machine} {Learning} {Tutorial} for {Operational} {Meteorology}. {Part} {I}},
url = {https://journals.ametsoc.org/view/journals/wefo/37/8/WAF-D-22-0070.1.xml},
doi = {10.1175/WAF-D-22-0070.1},
abstract = {Abstract
Recently, the use of machine learning in meteorology has increased greatly. While many machine learning methods are not new, university classes on machine learning are largely unavailable to meteorology students and are not required to become a meteorologist. The lack of formal instruction has contributed to perception that machine learning methods are “black boxes” and thus end-users are hesitant to apply the machine learning methods in their everyday workflow. To reduce the opaqueness of machine learning methods and lower hesitancy toward machine learning in meteorology, this paper provides a survey of some of the most common machine learning methods. A familiar meteorological example is used to contextualize the machine learning methods while also discussing machine learning topics using plain language. The following machine learning methods are demonstrated: linear regression, logistic regression, decision trees, random forest, gradient boosted decision trees, naïve Bayes, and support vector machines. Beyond discussing the different methods, the paper also contains discussions on the general machine learning process as well as best practices to enable readers to apply machine learning to their own datasets. Furthermore, all code (in the form of Jupyter notebooks and Google Colaboratory notebooks) used to make the examples in the paper is provided in an effort to catalyze the use of machine learning in meteorology.},
number = {8},
urldate = {2023-04-30},
journal = {Weather and Forecasting},
author = {Chase, Randy J. and Harrison, David R. and Burke, Amanda and Lackmann, Gary M. and McGovern, Amy},
month = aug,
year = {2022},
keywords = {ML, Machine Learning},
pages = {1509--1529},
}Reinforcement Learning in Manufacturing Control: Baselines, challenges and ways forward. Samsonov, V., Ben Hicham, K., & Meisen, T. Engineering Applications of Artificial Intelligence, 112:104868, June, 2022.
Paper doi bibtex
Paper doi bibtex @article{samsonov_reinforcement_2022,
title = {Reinforcement {Learning} in {Manufacturing} {Control}: {Baselines}, challenges and ways forward},
volume = {112},
issn = {09521976},
shorttitle = {Reinforcement {Learning} in {Manufacturing} {Control}},
url = {https://linkinghub.elsevier.com/retrieve/pii/S0952197622001130},
doi = {10.1016/j.engappai.2022.104868},
language = {en},
urldate = {2022-05-08},
journal = {Engineering Applications of Artificial Intelligence},
author = {Samsonov, Vladimir and Ben Hicham, Karim and Meisen, Tobias},
month = jun,
year = {2022},
keywords = {machine learning, mentions sympy, reinforcement learning},
pages = {104868},
}SYMBA: Symbolic Computation of Squared Amplitudes in High Energy Physics with Machine Learning. Alnuqaydan, A., Gleyzer, S., & Prosper, H. June, 2022. arXiv:2206.08901 [hep-ph]
Paper abstract bibtex
Paper abstract bibtex The cross section is one of the most important physical quantities in high-energy physics and the most time consuming to compute. While machine learning has proven to be highly successful in numerical calculations in high-energy physics, analytical calculations using machine learning are still in their infancy. In this work, we use a sequence-to-sequence transformer model to compute a key element of the cross section calculation, namely, the squared amplitude of an interaction. We show that a transformer model is able to predict correctly 89.0% and 99.4% of squared amplitudes of QCD and QED processes, respectively. We discuss the performance of the current model, its limitations and possible future directions for this work.
@misc{alnuqaydan_symba_2022,
title = {{SYMBA}: {Symbolic} {Computation} of {Squared} {Amplitudes} in {High} {Energy} {Physics} with {Machine} {Learning}},
shorttitle = {{SYMBA}},
url = {http://arxiv.org/abs/2206.08901},
abstract = {The cross section is one of the most important physical quantities in high-energy physics and the most time consuming to compute. While machine learning has proven to be highly successful in numerical calculations in high-energy physics, analytical calculations using machine learning are still in their infancy. In this work, we use a sequence-to-sequence transformer model to compute a key element of the cross section calculation, namely, the squared amplitude of an interaction. We show that a transformer model is able to predict correctly 89.0\% and 99.4\% of squared amplitudes of QCD and QED processes, respectively. We discuss the performance of the current model, its limitations and possible future directions for this work.},
urldate = {2022-07-04},
publisher = {arXiv},
author = {Alnuqaydan, Abdulhakim and Gleyzer, Sergei and Prosper, Harrison},
month = jun,
year = {2022},
note = {arXiv:2206.08901 [hep-ph]},
keywords = {high energy physics, machine learning, mentions sympy},
}We need to talk about random seeds. Bethard, S. October, 2022.
Paper doi bibtex 5 downloads
Paper doi bibtex 5 downloads @misc{https://doi.org/10.48550/arxiv.2210.13393,
doi = {10.48550/ARXIV.2210.13393},
url = {https://arxiv.org/abs/2210.13393},
author = {Bethard, Steven},
keywords = {machine learning},
title = {We need to talk about random seeds},
organization = {arXiv},
year = {2022},
month = oct,
}Symmetry Group Equivariant Architectures for Physics. Bogatskiy, A., Ganguly, S., Kipf, T., Kondor, R., Miller, D. W., Murnane, D., Offermann, J. T., Pettee, M., Shanahan, P., Shimmin, C., & Thais, S. arXiv:2203.06153 [astro-ph, physics:hep-ex, physics:hep-ph], March, 2022. arXiv: 2203.06153
Paper abstract bibtex
Paper abstract bibtex Physical theories grounded in mathematical symmetries are an essential component of our understanding of a wide range of properties of the universe. Similarly, in the domain of machine learning, an awareness of symmetries such as rotation or permutation invariance has driven impressive performance breakthroughs in computer vision, natural language processing, and other important applications. In this report, we argue that both the physics community and the broader machine learning community have much to understand and potentially to gain from a deeper investment in research concerning symmetry group equivariant machine learning architectures. For some applications, the introduction of symmetries into the fundamental structural design can yield models that are more economical (i.e. contain fewer, but more expressive, learned parameters), interpretable (i.e. more explainable or directly mappable to physical quantities), and/or trainable (i.e. more efficient in both data and computational requirements). We discuss various figures of merit for evaluating these models as well as some potential benefits and limitations of these methods for a variety of physics applications. Research and investment into these approaches will lay the foundation for future architectures that are potentially more robust under new computational paradigms and will provide a richer description of the physical systems to which they are applied.
@article{bogatskiy_symmetry_2022,
title = {Symmetry {Group} {Equivariant} {Architectures} for {Physics}},
url = {http://arxiv.org/abs/2203.06153},
abstract = {Physical theories grounded in mathematical symmetries are an essential component of our understanding of a wide range of properties of the universe. Similarly, in the domain of machine learning, an awareness of symmetries such as rotation or permutation invariance has driven impressive performance breakthroughs in computer vision, natural language processing, and other important applications. In this report, we argue that both the physics community and the broader machine learning community have much to understand and potentially to gain from a deeper investment in research concerning symmetry group equivariant machine learning architectures. For some applications, the introduction of symmetries into the fundamental structural design can yield models that are more economical (i.e. contain fewer, but more expressive, learned parameters), interpretable (i.e. more explainable or directly mappable to physical quantities), and/or trainable (i.e. more efficient in both data and computational requirements). We discuss various figures of merit for evaluating these models as well as some potential benefits and limitations of these methods for a variety of physics applications. Research and investment into these approaches will lay the foundation for future architectures that are potentially more robust under new computational paradigms and will provide a richer description of the physical systems to which they are applied.},
urldate = {2022-03-19},
journal = {arXiv:2203.06153 [astro-ph, physics:hep-ex, physics:hep-ph]},
author = {Bogatskiy, Alexander and Ganguly, Sanmay and Kipf, Thomas and Kondor, Risi and Miller, David W. and Murnane, Daniel and Offermann, Jan T. and Pettee, Mariel and Shanahan, Phiala and Shimmin, Chase and Thais, Savannah},
month = mar,
year = {2022},
note = {arXiv: 2203.06153},
keywords = {astrophysics, high energy physics, machine learning, mentions sympy},
}The Prediction and Influential Factors of Violence in Male Schizophrenia Patients With Machine Learning Algorithms. Yu, T., Zhang, X., Liu, X., Xu, C., & Deng, C. Frontiers in Psychiatry, 13(79989):9, 2022.
Paper doi abstract bibtex
Paper doi abstract bibtex Background: Early to identify male schizophrenia patients with violence is important for the performance of targeted measures and closer monitoring, but it is difficult to use conventional risk factors. This study is aimed to employ machine learning (ML) algorithms combined with routine data to predict violent behavior among male schizophrenia patients. Moreover, the identified best model might be utilized to calculate the probability of an individual committing violence. Method: We enrolled a total of 397 male schizophrenia patients and randomly stratified them into the training set and the testing set, in a 7:3 ratio. We used eight ML algorithms to develop the predictive models. The main variables as input features selected by the least absolute shrinkage and selection operator (LASSO) and logistic regression (LR) were integrated into prediction models for violence among male schizophrenia patients. In the training set, 10 × 10-fold cross-validation was conducted to adjust the parameters. In the testing set, we evaluated and compared the predictive performance of eight ML algorithms in terms of area under the curve (AUC) for the receiver operating characteristic curve. Result: Our results showed the prevalence of violence among male schizophrenia patients was 36.8%. The LASSO and LR identified main risk factors for violent behavior in patients with schizophrenia integrated into the predictive models, including lower education level [0.556 (0.378–0.816)], having cigarette smoking [2.121 (1.191–3.779)], higher positive syndrome [1.016 (1.002–1.031)] and higher social disability screening schedule (SDSS) [1.081 (1.026–1.139)]. The Neural Net (nnet) with an AUC of 0.6673 (0.5599–0.7748) had better prediction ability than that of other algorithms. Conclusion: ML algorithms are useful in early identifying male schizophrenia patients with violence and helping clinicians take preventive measures.
@article{yzlxd22,
abstract = {Background: Early to identify male schizophrenia patients with violence is important for the performance of targeted measures and closer monitoring, but it is difficult to use conventional risk factors. This study is aimed to employ machine learning (ML) algorithms combined with routine data to predict violent behavior among male schizophrenia patients. Moreover, the identified best model might be utilized to calculate the probability of an individual committing violence. Method: We enrolled a total of 397 male schizophrenia patients and randomly stratified them into the training set and the testing set, in a 7:3 ratio. We used eight ML algorithms to develop the predictive models. The main variables as input features selected by the least absolute shrinkage and selection operator (LASSO) and logistic regression (LR) were integrated into prediction models for violence among male schizophrenia patients. In the training set, 10 × 10-fold cross-validation was conducted to adjust the parameters. In the testing set, we evaluated and compared the predictive performance of eight ML algorithms in terms of area under the curve (AUC) for the receiver operating characteristic curve. Result: Our results showed the prevalence of violence among male schizophrenia patients was 36.8%. The LASSO and LR identified main risk factors for violent behavior in patients with schizophrenia integrated into the predictive models, including lower education level [0.556 (0.378–0.816)], having cigarette smoking [2.121 (1.191–3.779)], higher positive syndrome [1.016 (1.002–1.031)] and higher social disability screening schedule (SDSS) [1.081 (1.026–1.139)]. The Neural Net (nnet) with an AUC of 0.6673 (0.5599–0.7748) had better prediction ability than that of other algorithms. Conclusion: ML algorithms are useful in early identifying male schizophrenia patients with violence and helping clinicians take preventive measures.},
author = {Yu, Tao and Zhang, Xulai and Liu, Xiuyan and Xu, Chunyuan and Deng, Chenchen},
doi = {10.3389/fpsyt.2022.799899},
issn = {16640640},
journal = {Frontiers in Psychiatry},
keywords = {factor,machine learning,male,schizophrenia,violence},
number = {79989},
pages = {9},
title = {{The Prediction and Influential Factors of Violence in Male Schizophrenia Patients With Machine Learning Algorithms}},
url = {https://doi.org/10.3389/fpsyt.2022.799899},
volume = {13},
year = {2022}
}Curating Cyberbullying Datasets: a Human-AI Collaborative Approach. Gomez, C. E., Sztainberg, M. O., & Trana, R. E. International Journal of Bullying Prevention, 4(1):35–46, 2022.
Paper doi abstract bibtex
Paper doi abstract bibtex Cyberbullying is the use of digital communication tools and spaces to inflict physical, mental, or emotional distress. This serious form of aggression is frequently targeted at, but not limited to, vulnerable populations. A common problem when creating machine learning models to identify cyberbullying is the availability of accurately annotated, reliable, relevant, and diverse datasets. Datasets intended to train models for cyberbullying detection are typically annotated by human participants, which can introduce the following issues: (1) annotator bias, (2) incorrect annotation due to language and cultural barriers, and (3) the inherent subjectivity of the task can naturally create multiple valid labels for a given comment. The result can be a potentially inadequate dataset with one or more of these overlapping issues. We propose two machine learning approaches to identify and filter unambiguous comments in a cyberbullying dataset of roughly 19,000 comments collected from YouTube that was initially annotated using Amazon Mechanical Turk (AMT). Using consensus filtering methods, comments were classified as unambiguous when an agreement occurred between the AMT workers' majority label and the unanimous algorithmic filtering label. Comments identified as unambiguous were extracted and used to curate new datasets. We then used an artificial neural network to test for performance on these datasets. Compared to the original dataset, the classifier exhibits a large improvement in performance on modified versions of the dataset and can yield insight into the type of data that is consistently classified as bullying or non-bullying. This annotation approach can be expanded from cyberbullying datasets onto any classification corpus that has a similar complexity in scope.
@article{gst22,
abstract = {Cyberbullying is the use of digital communication tools and spaces to inflict physical, mental, or emotional distress. This serious form of aggression is frequently targeted at, but not limited to, vulnerable populations. A common problem when creating machine learning models to identify cyberbullying is the availability of accurately annotated, reliable, relevant, and diverse datasets. Datasets intended to train models for cyberbullying detection are typically annotated by human participants, which can introduce the following issues: (1) annotator bias, (2) incorrect annotation due to language and cultural barriers, and (3) the inherent subjectivity of the task can naturally create multiple valid labels for a given comment. The result can be a potentially inadequate dataset with one or more of these overlapping issues. We propose two machine learning approaches to identify and filter unambiguous comments in a cyberbullying dataset of roughly 19,000 comments collected from YouTube that was initially annotated using Amazon Mechanical Turk (AMT). Using consensus filtering methods, comments were classified as unambiguous when an agreement occurred between the AMT workers' majority label and the unanimous algorithmic filtering label. Comments identified as unambiguous were extracted and used to curate new datasets. We then used an artificial neural network to test for performance on these datasets. Compared to the original dataset, the classifier exhibits a large improvement in performance on modified versions of the dataset and can yield insight into the type of data that is consistently classified as bullying or non-bullying. This annotation approach can be expanded from cyberbullying datasets onto any classification corpus that has a similar complexity in scope.},
author = {Gomez, Christopher E. and Sztainberg, Marcelo O. and Trana, Rachel E.},
doi = {10.1007/s42380-021-00114-6},
issn = {25233661},
journal = {International Journal of Bullying Prevention},
keywords = {Consensus filtering,Cyberbullying,Data annotation,Machine learning,Supervised learning,YouTube},
number = {1},
pages = {35--46},
title = {{Curating Cyberbullying Datasets: a Human-AI Collaborative Approach}},
url = {https://doi.org/10.1007/s42380-021-00114-6},
volume = {4},
year = {2022}
}Development of a Machine Learning-based Soft Sensor for an Oil Refinery’s Distillation Column. Ferreira, J., Pedemonte, M., & Torres, A. I. Computers & Chemical Engineering, March, 2022.
Paper doi abstract bibtex
Paper doi abstract bibtex In this paper, a machine learning framework based on Kaizen Programming is proposed for building a soft-sensor using real historical data from an oil refinery. The soft-sensor estimates the composition of C4 hydrocarbons in the distillate stream of a splitter column. Kaizen Programming is a novel technique that has shown excellent results for symbolic regression problems without requiring a priori selection of the functional bases. The framework follows three different steps: pre-processing of the data, model building with the selected algorithm, and ensemble of different good-fitting models. The final ensemble model accurately predicts operation of the column at and between two steady states, outperforming the model based on Gaussian Process in both validation and soft-sensor degradation scenarios.
@article{ferreira_development_2022,
title = {Development of a {Machine} {Learning}-based {Soft} {Sensor} for an {Oil} {Refinery}’s {Distillation} {Column}},
issn = {0098-1354},
url = {https://www.sciencedirect.com/science/article/pii/S0098135422000977},
doi = {10.1016/j.compchemeng.2022.107756},
abstract = {In this paper, a machine learning framework based on Kaizen Programming is proposed for building a soft-sensor using real historical data from an oil refinery. The soft-sensor estimates the composition of C4 hydrocarbons in the distillate stream of a splitter column. Kaizen Programming is a novel technique that has shown excellent results for symbolic regression problems without requiring a priori selection of the functional bases. The framework follows three different steps: pre-processing of the data, model building with the selected algorithm, and ensemble of different good-fitting models. The final ensemble model accurately predicts operation of the column at and between two steady states, outperforming the model based on Gaussian Process in both validation and soft-sensor degradation scenarios.},
language = {en},
urldate = {2022-03-16},
journal = {Computers \& Chemical Engineering},
author = {Ferreira, Jimena and Pedemonte, Martín and Torres, Ana Inés},
month = mar,
year = {2022},
keywords = {Kaizen Programming, Machine Learning, Modeling from real data, Soft sensor, Surrogate models, mentions sympy},
pages = {107756},
}Analysis of Common Prediction Models for a Fuzzy Connected Source Target Production Based on Time Dependent Significance. Soller, S., Hoelzl, G., Greiler, T., & Kranz, M. In Proceedings of the 2022 International Conference on Embedded Wireless Systems and Networks, of EWSN '22, pages 226–231, USA, 2022. Junction Publishing.
Paper abstract bibtex
Paper abstract bibtex In an industrial setup quality measurements are taken in multiple steps of each production chain. Often a single product is evaluated for several steps in the production, but even more often pooled tests are done, to evaluate the quality of the used material or a charge. Instead of the traditional classification of sensor measurement to quality on a single process step, the question arises, how these steps interact with each other. Is it possible to foresee the faults of later production steps, by analyzing data gathered earlier? Especially in mass production this leads to unclear and fuzzy relationships. The material quality might not be known for every work piece produced but controlled in a fixed time interval. The challenge of these processes is, to correctly connect ground truth to feature vector by their temporal connection. In this work we show multiple steps to reach a better classification and insight into the production process. We gather data from a real-world environment and as a first shot, use common machine learning methods, which are available through public libraries. Therefore, we created a simple connection between the material trace elements and quality inspection. Further data analysis suggested the influence of the exact time of the quality inspection related to the measurement of trace elements. We performed a significance test, to proof the difference of time groups to each other. The identical machine learning methods were applied to these time groups and an improvement of classification accuracy of 2% could be detected. For feature approaches we propose an automatic split system, to find time dependent groups inside the data and split the data accordingly.
@inproceedings{ewsn2022_1,
abstract = {In an industrial setup quality measurements are taken in multiple steps of each production chain. Often a single product is evaluated for several steps in the production, but even more often pooled tests are done, to evaluate the quality of the used material or a charge. Instead of the traditional classification of sensor measurement to quality on a single process step, the question arises, how these steps interact with each other. Is it possible to foresee the faults of later production steps, by analyzing data gathered earlier? Especially in mass production this leads to unclear and fuzzy relationships. The material quality might not be known for every work piece produced but controlled in a fixed time interval. The challenge of these processes is, to correctly connect ground truth to feature vector by their temporal connection. In this work we show multiple steps to reach a better classification and insight into the production process. We gather data from a real-world environment and as a first shot, use common machine learning methods, which are available through public libraries. Therefore, we created a simple connection between the material trace elements and quality inspection. Further data analysis suggested the influence of the exact time of the quality inspection related to the measurement of trace elements. We performed a significance test, to proof the difference of time groups to each other. The identical machine learning methods were applied to these time groups and an improvement of classification accuracy of 2\% could be detected. For feature approaches we propose an automatic split system, to find time dependent groups inside the data and split the data accordingly.},
address = {USA},
author = {Soller, Sebastian and Hoelzl, Gerold and Greiler, Tobias and Kranz, Matthias},
booktitle = {Proceedings of the 2022 International Conference on Embedded Wireless Systems and Networks},
date-added = {2023-02-03 09:18:58 +0100},
date-modified = {2023-02-03 10:01:39 +0100},
keywords = {Reliability Keywords Industry 40, Machine Learning, Performance, Maintenance Prediction},
location = {Linz, Austria},
numpages = {6},
pages = {226--231},
publisher = {Junction Publishing},
series = {EWSN '22},
title = {Analysis of Common Prediction Models for a Fuzzy Connected Source Target Production Based on Time Dependent Significance},
url = {/publications/2022_ewsn_Fuzzy.pdf},
year = {2022},
bdsk-url-1 = {/publications/2022_ewsn_Fuzzy.pdf}}Population heterogeneity in clinical cohorts affects the predictive accuracy of brain imaging. Benkarim, O., Paquola, C., Park, B., Kebets, V., Hong, S., Wael, R. V. d., Zhang, S., Yeo, B. T. T., Eickenberg, M., Ge, T., Poline, J., Bernhardt, B. C., & Bzdok, D. PLOS Biology, 20(4):e3001627, April, 2022. Publisher: Public Library of Science
Paper doi abstract bibtex
Paper doi abstract bibtex Brain imaging research enjoys increasing adoption of supervised machine learning for single-participant disease classification. Yet, the success of these algorithms likely depends on population diversity, including demographic differences and other factors that may be outside of primary scientific interest. Here, we capitalize on propensity scores as a composite confound index to quantify diversity due to major sources of population variation. We delineate the impact of population heterogeneity on the predictive accuracy and pattern stability in 2 separate clinical cohorts: the Autism Brain Imaging Data Exchange (ABIDE, n = 297) and the Healthy Brain Network (HBN, n = 551). Across various analysis scenarios, our results uncover the extent to which cross-validated prediction performances are interlocked with diversity. The instability of extracted brain patterns attributable to diversity is located preferentially in regions part of the default mode network. Collectively, our findings highlight the limitations of prevailing deconfounding practices in mitigating the full consequences of population diversity.
@article{benkarim_population_2022,
title = {Population heterogeneity in clinical cohorts affects the predictive accuracy of brain imaging},
volume = {20},
issn = {1545-7885},
url = {https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3001627},
doi = {10.1371/journal.pbio.3001627},
abstract = {Brain imaging research enjoys increasing adoption of supervised machine learning for single-participant disease classification. Yet, the success of these algorithms likely depends on population diversity, including demographic differences and other factors that may be outside of primary scientific interest. Here, we capitalize on propensity scores as a composite confound index to quantify diversity due to major sources of population variation. We delineate the impact of population heterogeneity on the predictive accuracy and pattern stability in 2 separate clinical cohorts: the Autism Brain Imaging Data Exchange (ABIDE, n = 297) and the Healthy Brain Network (HBN, n = 551). Across various analysis scenarios, our results uncover the extent to which cross-validated prediction performances are interlocked with diversity. The instability of extracted brain patterns attributable to diversity is located preferentially in regions part of the default mode network. Collectively, our findings highlight the limitations of prevailing deconfounding practices in mitigating the full consequences of population diversity.},
language = {en},
number = {4},
urldate = {2022-05-04},
journal = {PLOS Biology},
author = {Benkarim, Oualid and Paquola, Casey and Park, Bo-yong and Kebets, Valeria and Hong, Seok-Jun and Wael, Reinder Vos de and Zhang, Shaoshi and Yeo, B. T. Thomas and Eickenberg, Michael and Ge, Tian and Poline, Jean-Baptiste and Bernhardt, Boris C. and Bzdok, Danilo},
month = apr,
year = {2022},
note = {Publisher: Public Library of Science},
keywords = {ADHD, Autism, Autism spectrum disorder, Forecasting, Machine learning, Neural networks, Neuroimaging, Species diversity},
pages = {e3001627},
}2021 (9)
A reinforcement learning iterated local search for makespan minimization in additive manufacturing machine scheduling problems. Alicastro, M., Ferone, D., Festa, P., Fugaro, S., & Pastore, T. Computers and Operations Research, 131:105272, Pergamon, jul, 2021.
Paper doi abstract bibtex
Paper doi abstract bibtex Additive manufacturing – also known as 3D printing – is a manufacturing process that is attracting more and more interest due to high production rates and reduced costs. This paper focuses on the scheduling problem of multiple additive manufacturing machines, recently proposed in the scientific literature. Given its intractability, instances of relevant size of additive manufacturing (AM) machine scheduling problem cannot be solved in reasonable computational times through mathematical models. For this reason, this paper proposes a Reinforcement Learning Iterated Local Search meta-heuristic, based on the implementation of a Q-Learning Variable Neighborhood Search, to provide heuristically good solutions at the cost of low computational expenses. A comprehensive computational study is conducted, comparing the proposed methodology with the results achieved by the CPLEX solver and to the performance of an Evolutionary Algorithm recently proposed for a similar problem, and adapted for the AM machine scheduling problem. Additionally, to explore the trade-off between efficiency and effectiveness more deeply, we present a further set of experiments that test the potential inclusion of a probabilistic stopping rule. The numerical results evidence that the proposed Reinforcement Learning Iterated Local Search is able to obtain statistically significant improvements compared to the other solution approaches featured in the computational experiments.
@article{Alicastro2021,
abstract = {Additive manufacturing – also known as 3D printing – is a manufacturing process that is attracting more and more interest due to high production rates and reduced costs. This paper focuses on the scheduling problem of multiple additive manufacturing machines, recently proposed in the scientific literature. Given its intractability, instances of relevant size of additive manufacturing (AM) machine scheduling problem cannot be solved in reasonable computational times through mathematical models. For this reason, this paper proposes a Reinforcement Learning Iterated Local Search meta-heuristic, based on the implementation of a Q-Learning Variable Neighborhood Search, to provide heuristically good solutions at the cost of low computational expenses. A comprehensive computational study is conducted, comparing the proposed methodology with the results achieved by the CPLEX solver and to the performance of an Evolutionary Algorithm recently proposed for a similar problem, and adapted for the AM machine scheduling problem. Additionally, to explore the trade-off between efficiency and effectiveness more deeply, we present a further set of experiments that test the potential inclusion of a probabilistic stopping rule. The numerical results evidence that the proposed Reinforcement Learning Iterated Local Search is able to obtain statistically significant improvements compared to the other solution approaches featured in the computational experiments.},
author = {Alicastro, Mirko and Ferone, Daniele and Festa, Paola and Fugaro, Serena and Pastore, Tommaso},
doi = {10.1016/j.cor.2021.105272},
file = {:C\:/Users/optit/Documents/Mendeley Desktop/Alicastro et al. - 2021 - A reinforcement learning iterated local search for makespan minimization in additive manufacturing machine sch.pdf:pdf},
issn = {03050548},
journal = {Computers and Operations Research},
keywords = {Machine scheduling,Q-learning,additive manufacturing,iterated local search,machine learning,reinforcement learning,scheduling,vns},
mendeley-tags = {iterated local search,machine learning,reinforcement learning,scheduling,vns},
month = {jul},
pages = {105272},
publisher = {Pergamon},
title = {{A reinforcement learning iterated local search for makespan minimization in additive manufacturing machine scheduling problems}},
url = {https://linkinghub.elsevier.com/retrieve/pii/S0305054821000642},
volume = {131},
year = {2021}
}Deep learning for physics research. Erdmann, M. World Scientific, New Jersey, 2021.
abstract bibtex
abstract bibtex
"A core principle of physics is knowledge gained from data. Thus, deep learning has instantly entered physics and may become a new paradigm in basic and applied research. This textbook addresses physics students and physicists who want to understand what deep learning actually means, and what is the potential for their own scientific projects. Being familiar with linear algebra and parameter optimization is sufficient to jump-start deep learning. Adopting a pragmatic approach, basic and advanced applications in physics research are described. Also offered are simple hands-on exercises for implementing deep networks for which python code and training data can be downloaded"–
@book{erdmann_deep_2021,
address = {New Jersey},
title = {Deep learning for physics research},
isbn = {9789811237478 9789811237461},
abstract = {"A core principle of physics is knowledge gained from data. Thus, deep learning has instantly entered physics and may become a new paradigm in basic and applied research. This textbook addresses physics students and physicists who want to understand what deep learning actually means, and what is the potential for their own scientific projects. Being familiar with linear algebra and parameter optimization is sufficient to jump-start deep learning. Adopting a pragmatic approach, basic and advanced applications in physics research are described. Also offered are simple hands-on exercises for implementing deep networks for which python code and training data can be downloaded"--},
publisher = {World Scientific},
author = {Erdmann, Martin},
year = {2021},
keywords = {Data processing, Machine learning, Physics, Research},
}Use of Electronic Health Data for Disease Prediction: A Comprehensive Literature Review. Hossain, M. E., Khan, A., Moni, M. A., & Uddin, S. IEEE/ACM Transactions on Computational Biology and Bioinformatics, 18(2):745–758, March, 2021. Conference Name: IEEE/ACM Transactions on Computational Biology and Bioinformatics
doi abstract bibtex
doi abstract bibtex
Disease prediction has the potential to benefit stakeholders such as the government and health insurance companies. It can identify patients at risk of disease or health conditions. Clinicians can then take appropriate measures to avoid or minimize the risk and in turn, improve quality of care and avoid potential hospital admissions. Due to the recent advancement of tools and techniques for data analytics, disease risk prediction can leverage large amounts of semantic information, such as demographics, clinical diagnosis and measurements, health behaviours, laboratory results, prescriptions and care utilisation. In this regard, electronic health data can be a potential choice for developing disease prediction models. A significant number of such disease prediction models have been proposed in the literature over time utilizing large-scale electronic health databases, different methods, and healthcare variables. The goal of this comprehensive literature review was to discuss different risk prediction models that have been proposed based on electronic health data. Search terms were designed to find relevant research articles that utilized electronic health data to predict disease risks. Online scholarly databases were searched to retrieve results, which were then reviewed and compared in terms of the method used, disease type, and prediction accuracy. This paper provides a comprehensive review of the use of electronic health data for risk prediction models. A comparison of the results from different techniques for three frequently modelled diseases using electronic health data was also discussed in this study. In addition, the advantages and disadvantages of different risk prediction models, as well as their performance, were presented. Electronic health data have been widely used for disease prediction. A few modelling approaches show very high accuracy in predicting different diseases using such data. These modelling approaches have been used to inform the clinical decision process to achieve better outcomes.
@article{hossain_use_2021,
title = {Use of {Electronic} {Health} {Data} for {Disease} {Prediction}: {A} {Comprehensive} {Literature} {Review}},
volume = {18},
issn = {1557-9964},
shorttitle = {Use of {Electronic} {Health} {Data} for {Disease} {Prediction}},
doi = {10.1109/TCBB.2019.2937862},
abstract = {Disease prediction has the potential to benefit stakeholders such as the government and health insurance companies. It can identify patients at risk of disease or health conditions. Clinicians can then take appropriate measures to avoid or minimize the risk and in turn, improve quality of care and avoid potential hospital admissions. Due to the recent advancement of tools and techniques for data analytics, disease risk prediction can leverage large amounts of semantic information, such as demographics, clinical diagnosis and measurements, health behaviours, laboratory results, prescriptions and care utilisation. In this regard, electronic health data can be a potential choice for developing disease prediction models. A significant number of such disease prediction models have been proposed in the literature over time utilizing large-scale electronic health databases, different methods, and healthcare variables. The goal of this comprehensive literature review was to discuss different risk prediction models that have been proposed based on electronic health data. Search terms were designed to find relevant research articles that utilized electronic health data to predict disease risks. Online scholarly databases were searched to retrieve results, which were then reviewed and compared in terms of the method used, disease type, and prediction accuracy. This paper provides a comprehensive review of the use of electronic health data for risk prediction models. A comparison of the results from different techniques for three frequently modelled diseases using electronic health data was also discussed in this study. In addition, the advantages and disadvantages of different risk prediction models, as well as their performance, were presented. Electronic health data have been widely used for disease prediction. A few modelling approaches show very high accuracy in predicting different diseases using such data. These modelling approaches have been used to inform the clinical decision process to achieve better outcomes.},
number = {2},
journal = {IEEE/ACM Transactions on Computational Biology and Bioinformatics},
author = {Hossain, M. E. and Khan, A. and Moni, M. A. and Uddin, S.},
month = mar,
year = {2021},
note = {Conference Name: IEEE/ACM Transactions on Computational Biology and Bioinformatics},
keywords = {Computational modeling, Consumer electronics, Data models, Disease prediction, Diseases, Logistics, ML, Machine Learning, Predictive models, data mining, electronic health data, healthcare informatics, social network analysis},
pages = {745--758},
}Recurrent localization networks applied to the Lippmann-Schwinger equation. Kelly, C. & Kalidindi, S. R. Computational Materials Science, 2021.
bibtex
bibtex
@article{rlnpaper,
title = {Recurrent localization networks applied to the {Lippmann}-{Schwinger} equation},
journal = {Computational Materials Science},
year = {2021},
issn = {0927-0256},
author = {Conlain Kelly and Surya R. Kalidindi},
keywords = {Machine learning, Learned optimization, Localization, Convolutional neural networks}}Can deep learning beat numerical weather prediction?. Schultz, M. G., Betancourt, C., Gong, B., Kleinert, F., Langguth, M., Leufen, L. H., Mozaffari, A., & Stadtler, S. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 379(2194):20200097, April, 2021.
Paper doi abstract bibtex
Paper doi abstract bibtex The recent hype about artificial intelligence has sparked renewed interest in applying the successful deep learning (DL) methods for image recognition, speech recognition, robotics, strategic games and other application areas to the field of meteorology. There is some evidence that better weather forecasts can be produced by introducing big data mining and neural networks into the weather prediction workflow. Here, we discuss the question of whether it is possible to completely replace the current numerical weather models and data assimilation systems with DL approaches. This discussion entails a review of state-of-the-art machine learning concepts and their applicability to weather data with its pertinent statistical properties. We think that it is not inconceivable that numerical weather models may one day become obsolete, but a number of fundamental breakthroughs are needed before this goal comes into reach. This article is part of the theme issue ‘Machine learning for weather and climate modelling’.
@article{schultz_can_2021,
title = {Can deep learning beat numerical weather prediction?},
volume = {379},
issn = {1364-503X, 1471-2962},
url = {https://royalsocietypublishing.org/doi/10.1098/rsta.2020.0097},
doi = {10.1098/rsta.2020.0097},
abstract = {The recent hype about artificial intelligence has sparked renewed interest in applying the successful deep learning (DL) methods for image recognition, speech recognition, robotics, strategic games and other application areas to the field of meteorology. There is some evidence that better weather forecasts can be produced by introducing big data mining and neural networks into the weather prediction workflow. Here, we discuss the question of whether it is possible to completely replace the current numerical weather models and data assimilation systems with DL approaches. This discussion entails a review of state-of-the-art machine learning concepts and their applicability to weather data with its pertinent statistical properties. We think that it is not inconceivable that numerical weather models may one day become obsolete, but a number of fundamental breakthroughs are needed before this goal comes into reach.
This article is part of the theme issue ‘Machine learning for weather and climate modelling’.},
language = {en},
number = {2194},
urldate = {2021-11-15},
journal = {Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences},
author = {Schultz, M. G. and Betancourt, C. and Gong, B. and Kleinert, F. and Langguth, M. and Leufen, L. H. and Mozaffari, A. and Stadtler, S.},
month = apr,
year = {2021},
keywords = {AI, AI4EO, AIforEOWS, ML, Machine Learning, ml4esop},
pages = {20200097},
}Semantization of Machine Learning and Data Science (a Project Idea). Alexiev, V. & Boytcheva, S. presentation, September, 2021.
Paper abstract bibtex
Paper abstract bibtex Problem: Data Science, AI & ML are expensive, and that's one of the reasons why relatively few enterprises use them. Goal: rationalize and industrialize DS efforts, and make them more reproducible and reusable. Approach: capture a lot of semantic info about all DS processes in an enterprise, and thus enable automation, discovery, reusability. The kinds of data we'd like to represent and integrate semantically (part of it is similar to what you can see on the Kaggle and OpenML sites): - Business context: goals, motivations, data value, value chain, cost vs benefit analysis, SWOT analysis... - DS challenges, where do they come from, datasets that can be leveraged to solve them - DS staff, expertise, projects, tasks, risks - DS/ML algorithms, implementations, modules, dependencies, software projects, versions, issue trackers - Cloud and IT resources: compute, storage; their deployment, management, automation... - ML model deployment, performance, model drift, retraining… Established software genres that cover parts of this landscape: - ModelOps (devOps for ML), Feature Spaces - Enterprise data catalogs (data hubs) vs data marketplaces vs open data catalogs vs EU Data Spaces and their metadata - FAIR data, reproducible research, Research Objects, research workflows, We've researched over 100 relevant ontologies that can be leveraged, covering - Organizations/enterprises, business plans, - Ontologies, semantic data, - DS challenges, datasets, statistical data, quality assessment - DS/ML approaches, software, projects, issues, - Data on research/science - Project management Focusing on DS/ML approaches only, a couple of the relevant ontologies or standards are: - PMML (predictive modeling markup language) - e-LICO, DMEX ontologies for describing DS - OntoDM, KDO ontologies for describing DS
@Misc{AlexievBoytcheva2021-SemantizationML,
author = {Vladimir Alexiev and Svetla Boytcheva},
title = {{Semantization of Machine Learning and Data Science (a Project Idea)}},
howpublished = {presentation},
month = sep,
year = 2021,
url = {https://docs.google.com/presentation/d/1_8LSXa9vVzNwPE6Hjj4cKIJNRRBNz2wP/edit},
keywords = {Ontotext, research projects, knowledge graph, KG technologies, Semantization, Machine Learning, Data Science},
address = {Presentation at Big Dava Value Association Activity Group 45 (BDVA AG 45)},
abstract = {Problem: Data Science, AI & ML are expensive, and that's one of the reasons why relatively few enterprises use them.
Goal: rationalize and industrialize DS efforts, and make them more reproducible and reusable.
Approach: capture a lot of semantic info about all DS processes in an enterprise, and thus enable automation, discovery, reusability.
The kinds of data we'd like to represent and integrate semantically (part of it is similar to what you can see on the Kaggle and OpenML sites):
- Business context: goals, motivations, data value, value chain, cost vs benefit analysis, SWOT analysis...
- DS challenges, where do they come from, datasets that can be leveraged to solve them
- DS staff, expertise, projects, tasks, risks
- DS/ML algorithms, implementations, modules, dependencies, software projects, versions, issue trackers
- Cloud and IT resources: compute, storage; their deployment, management, automation...
- ML model deployment, performance, model drift, retraining…
Established software genres that cover parts of this landscape:
- ModelOps (devOps for ML), Feature Spaces
- Enterprise data catalogs (data hubs) vs data marketplaces vs open data catalogs vs EU Data Spaces and their metadata
- FAIR data, reproducible research, Research Objects, research workflows,
We've researched over 100 relevant ontologies that can be leveraged, covering
- Organizations/enterprises, business plans,
- Ontologies, semantic data,
- DS challenges, datasets, statistical data, quality assessment
- DS/ML approaches, software, projects, issues,
- Data on research/science
- Project management
Focusing on DS/ML approaches only, a couple of the relevant ontologies or standards are:
- PMML (predictive modeling markup language)
- e-LICO, DMEX ontologies for describing DS
- OntoDM, KDO ontologies for describing DS},
}Fault Detection in DC Microgrids Using Short-Time Fourier Transform. Grcić, I., Pandžić, H., & Novosel, D. Energies, 14(2):14, 2021.
doi bibtex 2 downloads
doi bibtex 2 downloads
@article { 10.3390/en14020277,
author = {Grci\'{c}, Ivan and Pand\v{z}i\'{c}, Hrvoje and Novosel, Damir},
year = {2021},
pages = {14},
DOI = {10.3390/en14020277},
chapter = {277},
keywords = {short-time Fourier transform, intelligent classifiers, microgrid, fault detection, machine learning},
journal = {Energies},
doi = {10.3390/en14020277},
volume = {14},
number = {2},
title = {Fault Detection in DC Microgrids Using Short-Time
Fourier Transform},
keyword = {short-time Fourier transform, intelligent classifiers, microgrid, fault detection, machine learning},
chapternumber = {277}
}A Systematic Review of Artificial Intelligence Public Datasets for Railway Applications. Pappaterra, M. J., Flammini, F., Vittorini, V., & Bešinović, N. Infrastructures, 6(10):136, October, 2021. Number: 10 Publisher: Multidisciplinary Digital Publishing Institute
Paper doi abstract bibtex
Paper doi abstract bibtex The aim of this paper is to review existing publicly available and open artificial intelligence (AI) oriented datasets in different domains and subdomains of the railway sector. The contribution of this paper is an overview of AI-oriented railway data published under Creative Commons (CC) or any other copyright type that entails public availability and freedom of use. These data are of great value for open research and publications related to the application of AI in the railway sector. This paper includes insights on the public railway data: we distinguish different subdomains, including maintenance and inspection, traffic planning and management, safety and security and type of data including numerical, string, image and other. The datasets reviewed cover the last three decades, from January 1990 to January 2021. The study revealed that the number of open datasets is very small in comparison with the available literature related to AI applications in the railway industry. Another shortcoming is the lack of documentation and metadata on public datasets, including information related to missing data, collection schemes and other limitations. This study also presents quantitative data, such as the number of available open datasets divided by railway application, type of data and year of publication. This review also reveals that there are openly available APIs—maintained by government organizations and train operating companies (TOCs)—that can be of great use for data harvesting and can facilitate the creation of large public datasets. These data are usually well-curated real-time data that can greatly contribute to the accuracy of AI models. Furthermore, we conclude that the extension of AI applications in the railway sector merits a centralized hub for publicly available datasets and open APIs.
@article{pappaterra_systematic_2021,
title = {A {Systematic} {Review} of {Artificial} {Intelligence} {Public} {Datasets} for {Railway} {Applications}},
volume = {6},
copyright = {http://creativecommons.org/licenses/by/3.0/},
issn = {2412-3811},
url = {https://www.mdpi.com/2412-3811/6/10/136},
doi = {10.3390/infrastructures6100136},
abstract = {The aim of this paper is to review existing publicly available and open artificial intelligence (AI) oriented datasets in different domains and subdomains of the railway sector. The contribution of this paper is an overview of AI-oriented railway data published under Creative Commons (CC) or any other copyright type that entails public availability and freedom of use. These data are of great value for open research and publications related to the application of AI in the railway sector. This paper includes insights on the public railway data: we distinguish different subdomains, including maintenance and inspection, traffic planning and management, safety and security and type of data including numerical, string, image and other. The datasets reviewed cover the last three decades, from January 1990 to January 2021. The study revealed that the number of open datasets is very small in comparison with the available literature related to AI applications in the railway industry. Another shortcoming is the lack of documentation and metadata on public datasets, including information related to missing data, collection schemes and other limitations. This study also presents quantitative data, such as the number of available open datasets divided by railway application, type of data and year of publication. This review also reveals that there are openly available APIs—maintained by government organizations and train operating companies (TOCs)—that can be of great use for data harvesting and can facilitate the creation of large public datasets. These data are usually well-curated real-time data that can greatly contribute to the accuracy of AI models. Furthermore, we conclude that the extension of AI applications in the railway sector merits a centralized hub for publicly available datasets and open APIs.},
language = {en},
number = {10},
urldate = {2022-05-15},
journal = {Infrastructures},
author = {Pappaterra, Mauro José and Flammini, Francesco and Vittorini, Valeria and Bešinović, Nikola},
month = oct,
year = {2021},
note = {Number: 10
Publisher: Multidisciplinary Digital Publishing Institute},
keywords = {intelligent transportation, machine learning, predictive maintenance, public datasets, railways},
pages = {136},
}BridgeClassics. Künstliche Intelligenz für die Klassische Philologie. Beyer, A., Schulz, K., & Cordes, L. May, 2021.
Paper doi abstract bibtex 5 downloads
Paper doi abstract bibtex 5 downloads Folien zu einem virtuellen Vortrag der Humboldt-Universität zu Berlin in Kooperation mit der Technischen Universität Berlin
@misc{beyer_bridgeclassics_2021,
address = {online},
title = {{BridgeClassics}. {Künstliche} {Intelligenz} für die {Klassische} {Philologie}},
url = {https://zenodo.org/record/4745781#.YJkPT4MzbJw},
abstract = {Folien zu einem virtuellen Vortrag der Humboldt-Universität zu Berlin in Kooperation mit der Technischen Universität Berlin},
language = {deu},
urldate = {2021-05-10},
author = {Beyer, Andrea and Schulz, Konstantin and Cordes, Lisa},
month = may,
year = {2021},
doi = {10.5281/zenodo.4745781},
keywords = {Ancient Greek, Latin, artificial intelligence, classics, machine learning},
}2020 (10)
Deep Learning in Mining Biological Data. Mahmud, M., Kaiser, M. S., & Hussain, A. arXiv:2003.00108 [cs, q-bio, stat], February, 2020. arXiv: 2003.00108
Paper abstract bibtex
Paper abstract bibtex Recent technological advancements in data acquisition tools allowed life scientists to acquire multimodal data from different biological application domains. Broadly categorized in three types (i.e., sequences, images, and signals), these data are huge in amount and complex in nature. Mining such an enormous amount of data for pattern recognition is a big challenge and requires sophisticated data-intensive machine learning techniques. Artificial neural network-based learning systems are well known for their pattern recognition capabilities and lately their deep architectures - known as deep learning (DL) - have been successfully applied to solve many complex pattern recognition problems. Highlighting the role of DL in recognizing patterns in biological data, this article provides - applications of DL to biological sequences, images, and signals data; overview of open access sources of these data; description of open source DL tools applicable on these data; and comparison of these tools from qualitative and quantitative perspectives. At the end, it outlines some open research challenges in mining biological data and puts forward a number of possible future perspectives.
@article{mahmud_deep_2020,
title = {Deep {Learning} in {Mining} {Biological} {Data}},
url = {http://arxiv.org/abs/2003.00108},
abstract = {Recent technological advancements in data acquisition tools allowed life scientists to acquire multimodal data from different biological application domains. Broadly categorized in three types (i.e., sequences, images, and signals), these data are huge in amount and complex in nature. Mining such an enormous amount of data for pattern recognition is a big challenge and requires sophisticated data-intensive machine learning techniques. Artificial neural network-based learning systems are well known for their pattern recognition capabilities and lately their deep architectures - known as deep learning (DL) - have been successfully applied to solve many complex pattern recognition problems. Highlighting the role of DL in recognizing patterns in biological data, this article provides - applications of DL to biological sequences, images, and signals data; overview of open access sources of these data; description of open source DL tools applicable on these data; and comparison of these tools from qualitative and quantitative perspectives. At the end, it outlines some open research challenges in mining biological data and puts forward a number of possible future perspectives.},
urldate = {2020-03-07},
journal = {arXiv:2003.00108 [cs, q-bio, stat]},
author = {Mahmud, Mufti and Kaiser, M. Shamim and Hussain, Amir},
month = feb,
year = {2020},
note = {arXiv: 2003.00108},
keywords = {Quantitative Biology - Quantitative Methods, machine learning, mentions sympy},
}Prediction of physical violence in schizophrenia with machine learning algorithms. Wang, K. Z., Bani-Fatemi, A., Adanty, C., Harripaul, R., Griffiths, J., Kolla, N., Gerretsen, P., Graff, A., & De Luca, V. Psychiatry Research, 289:11296, 2020.
Paper doi abstract bibtex
Paper doi abstract bibtex Patients with schizophrenia have been shown to have an increased risk for physical violence. While certain features have been identified as risk factors, it has been difficult to integrate these variables to identify violent patients. The present study thus attempts to develop a clinically-relevant predictive tool. In a population of 275 schizophrenia patients, we identified 103 participants as violent and 172 as non-violent through electronic medical documentation, and conducted cross-sectional assessments to identify demographic, clinical, and sociocultural variables. Using these predictors, we utilized seven machine learning classification algorithms to predict for past instances of physical violence. Our classification algorithms predicted with significant accuracy compared to random discrimination alone, and had varying degrees of predictive power, as described by various performance measures. We determined that the random forest model performed marginally better than other algorithms, with an accuracy of 62% and an area under the receiver operator characteristic curve (AUROC) of 0.63. To summarize, machine learning classification algorithms are becoming increasingly valuable, though, optimization of these models is needed to better complement diagnostic decisions regarding early interventional measures to predict instances of physical violence.
@article{wbahgkggd20,
abstract = {Patients with schizophrenia have been shown to have an increased risk for physical violence. While certain features have been identified as risk factors, it has been difficult to integrate these variables to identify violent patients. The present study thus attempts to develop a clinically-relevant predictive tool. In a population of 275 schizophrenia patients, we identified 103 participants as violent and 172 as non-violent through electronic medical documentation, and conducted cross-sectional assessments to identify demographic, clinical, and sociocultural variables. Using these predictors, we utilized seven machine learning classification algorithms to predict for past instances of physical violence. Our classification algorithms predicted with significant accuracy compared to random discrimination alone, and had varying degrees of predictive power, as described by various performance measures. We determined that the random forest model performed marginally better than other algorithms, with an accuracy of 62% and an area under the receiver operator characteristic curve (AUROC) of 0.63. To summarize, machine learning classification algorithms are becoming increasingly valuable, though, optimization of these models is needed to better complement diagnostic decisions regarding early interventional measures to predict instances of physical violence.},
author = {Wang, Kevin Z. and Bani-Fatemi, Ali and Adanty, Christopher and Harripaul, Ricardo and Griffiths, John and Kolla, Nathan and Gerretsen, Philip and Graff, Ariel and {De Luca}, Vincenzo},
doi = {10.1016/j.psychres.2020.112960},
issn = {18727123},
journal = {Psychiatry Research},
keywords = {Childhood trauma,Machine learning,Personality,Schizophrenia,Violence},
pages = {11296},
pmid = {32361562},
title = {{Prediction of physical violence in schizophrenia with machine learning algorithms}},
url = {https://doi.org/10.1016/j.psychres.2020.112960},
volume = {289},
year = {2020}
}Finding warning markers: Leveraging natural language processing and machine learning technologies to detect risk of school violence. Ni, Y., Barzman, D., Bachtel, A., Griffey, M., Osborn, A., & Sorter, M. International Journal of Medical Informatics, 139(10413):7, 2020.
Paper doi abstract bibtex
Paper doi abstract bibtex Introduction: School violence has a far-reaching effect, impacting the entire school population including staff, students and their families. Among youth attending the most violent schools, studies have reported higher dropout rates, poor school attendance, and poor scholastic achievement. It was noted that the largest crime-prevention results occurred when youth at elevated risk were given an individualized prevention program. However, much work is needed to establish an effective approach to identify at-risk subjects. Objective: In our earlier research, we developed a risk assessment program to interview subjects, identify risk and protective factors, and evaluate risk for school violence. This study focused on developing natural language processing (NLP) and machine learning technologies to automate the risk assessment process. Material and methods: We prospectively recruited 131 students with or without behavioral concerns from 89 schools between 05/01/2015 and 04/30/2018. The subjects were interviewed with two risk assessment scales and a questionnaire, and their risk of violence were determined by pediatric psychiatrists based on clinical judgment. Using NLP technologies, different types of linguistic features were extracted from the interview content. Machine learning classifiers were then applied to predict risk of school violence for individual subjects. A two-stage feature selection was implemented to identify violence-related predictors. The performance was validated on the psychiatrist-generated reference standard of risk levels, where positive predictive value (PPV), sensitivity (SEN), negative predictive value (NPV), specificity (SPEC) and area under the ROC curve (AUC) were assessed. Results: Compared to subjects' sociodemographic information, use of linguistic features significantly improved classifiers' predictive performance (P < 0.01). The best-performing classifier with n-gram features achieved 86.5 %/86.5 %/85.7 %/85.7 %/94.0 % (PPV/SEN/NPV/SPEC/AUC) on the cross-validation set and 83.3 %/93.8 %/91.7 %/78.6 %/94.6 % (PPV/SEN/NPV/SPEC/AUC) on the test data. The feature selection process identified a set of predictors covering the discussion of subjects' thoughts, perspectives, behaviors, individual characteristics, peers and family dynamics, and protective factors. Conclusions: By analyzing the content from subject interviews, the NLP and machine learning algorithms showed good capacity for detecting risk of school violence. The feature selection uncovered multiple warning markers that could deliver useful clinical insights to assist personalizing intervention. Consequently, the developed approach offered the promise of an accurate and scalable computerized screening service for preventing school violence.
@article{nbbgos20,
abstract = {Introduction: School violence has a far-reaching effect, impacting the entire school population including staff, students and their families. Among youth attending the most violent schools, studies have reported higher dropout rates, poor school attendance, and poor scholastic achievement. It was noted that the largest crime-prevention results occurred when youth at elevated risk were given an individualized prevention program. However, much work is needed to establish an effective approach to identify at-risk subjects. Objective: In our earlier research, we developed a risk assessment program to interview subjects, identify risk and protective factors, and evaluate risk for school violence. This study focused on developing natural language processing (NLP) and machine learning technologies to automate the risk assessment process. Material and methods: We prospectively recruited 131 students with or without behavioral concerns from 89 schools between 05/01/2015 and 04/30/2018. The subjects were interviewed with two risk assessment scales and a questionnaire, and their risk of violence were determined by pediatric psychiatrists based on clinical judgment. Using NLP technologies, different types of linguistic features were extracted from the interview content. Machine learning classifiers were then applied to predict risk of school violence for individual subjects. A two-stage feature selection was implemented to identify violence-related predictors. The performance was validated on the psychiatrist-generated reference standard of risk levels, where positive predictive value (PPV), sensitivity (SEN), negative predictive value (NPV), specificity (SPEC) and area under the ROC curve (AUC) were assessed. Results: Compared to subjects' sociodemographic information, use of linguistic features significantly improved classifiers' predictive performance (P < 0.01). The best-performing classifier with n-gram features achieved 86.5 %/86.5 %/85.7 %/85.7 %/94.0 % (PPV/SEN/NPV/SPEC/AUC) on the cross-validation set and 83.3 %/93.8 %/91.7 %/78.6 %/94.6 % (PPV/SEN/NPV/SPEC/AUC) on the test data. The feature selection process identified a set of predictors covering the discussion of subjects' thoughts, perspectives, behaviors, individual characteristics, peers and family dynamics, and protective factors. Conclusions: By analyzing the content from subject interviews, the NLP and machine learning algorithms showed good capacity for detecting risk of school violence. The feature selection uncovered multiple warning markers that could deliver useful clinical insights to assist personalizing intervention. Consequently, the developed approach offered the promise of an accurate and scalable computerized screening service for preventing school violence.},
author = {Ni, Yizhao and Barzman, Drew and Bachtel, Alycia and Griffey, Marcus and Osborn, Alexander and Sorter, Michael},
doi = {10.1016/j.ijmedinf.2020.104137},
issn = {18728243},
journal = {International Journal of Medical Informatics},
keywords = {Automated risk assessment,Machine learning,Natural language processing,School violence},
number = {10413},
pages = {7},
pmid = {32361146},
title = {{Finding warning markers: Leveraging natural language processing and machine learning technologies to detect risk of school violence}},
url = {https://doi.org/10.1016/j.ijmedinf.2020.104137},
volume = {139},
year = {2020}
}Neural Identification for Control. Saha, P. & Mukhopadhyay, S. arXiv:2009.11782 [cs, eess], September, 2020. arXiv: 2009.11782
Paper abstract bibtex
Paper abstract bibtex We present a new method for learning control law that stabilizes an unknown nonlinear dynamical system at an equilibrium point. We formulate a system identification task in a self-supervised learning setting that jointly learns a controller and corresponding stable closed-loop dynamics hypothesis. The open-loop input-output behavior of the underlying dynamical system is used as the supervising signal to train the neural network-based system model and controller. The method relies on the Lyapunov stability theory to generate a stable closed-loop dynamics hypothesis and corresponding control law. We demonstrate our method on various nonlinear control problems such as n-Link pendulum balancing, pendulum on cart balancing, and wheeled vehicle path following.
@article{saha_neural_2020,
title = {Neural {Identification} for {Control}},
url = {http://arxiv.org/abs/2009.11782},
abstract = {We present a new method for learning control law that stabilizes an unknown nonlinear dynamical system at an equilibrium point. We formulate a system identification task in a self-supervised learning setting that jointly learns a controller and corresponding stable closed-loop dynamics hypothesis. The open-loop input-output behavior of the underlying dynamical system is used as the supervising signal to train the neural network-based system model and controller. The method relies on the Lyapunov stability theory to generate a stable closed-loop dynamics hypothesis and corresponding control law. We demonstrate our method on various nonlinear control problems such as n-Link pendulum balancing, pendulum on cart balancing, and wheeled vehicle path following.},
urldate = {2020-09-28},
journal = {arXiv:2009.11782 [cs, eess]},
author = {Saha, Priyabrata and Mukhopadhyay, Saibal},
month = sep,
year = {2020},
note = {arXiv: 2009.11782},
keywords = {machine learning, mentions sympy, robotics},
}An enhanced deep learning approach for brain cancer MRI images classification using residual networks. Ismael, S. A. [., Mohammed, A., & Hefny, H. Artificial Intelligence in Medicine, 102:101779, 2020.
Paper doi abstract bibtex
Paper doi abstract bibtex Cancer is the second leading cause of death after cardiovascular diseases. Out of all types of cancer, brain cancer has the lowest survival rate. Brain tumors can have different types depending on their shape, texture, and location. Proper diagnosis of the tumor type enables the doctor to make the correct treatment choice and help save the patient's life. There is a high need in the Artificial Intelligence field for a Computer Assisted Diagnosis (CAD) system to assist doctors and radiologists with the diagnosis and classification of tumors. Over recent years, deep learning has shown an optimistic performance in computer vision systems. In this paper, we propose an enhanced approach for classifying brain tumor types using Residual Networks. We evaluate the proposed model on a benchmark dataset containing 3064 MRI images of 3 brain tumor types (Meningiomas, Gliomas, and Pituitary tumors). We have achieved the highest accuracy of 99% outperforming the other previous work on the same dataset.
@article{ismael_enhanced_2020,
title = {An enhanced deep learning approach for brain cancer {MRI} images classification using residual networks},
volume = {102},
issn = {0933-3657},
url = {http://www.sciencedirect.com/science/article/pii/S0933365719306177},
doi = {https://doi.org/10.1016/j.artmed.2019.101779},
abstract = {Cancer is the second leading cause of death after cardiovascular diseases. Out of all types of cancer, brain cancer has the lowest survival rate. Brain tumors can have different types depending on their shape, texture, and location. Proper diagnosis of the tumor type enables the doctor to make the correct treatment choice and help save the patient's life. There is a high need in the Artificial Intelligence field for a Computer Assisted Diagnosis (CAD) system to assist doctors and radiologists with the diagnosis and classification of tumors. Over recent years, deep learning has shown an optimistic performance in computer vision systems. In this paper, we propose an enhanced approach for classifying brain tumor types using Residual Networks. We evaluate the proposed model on a benchmark dataset containing 3064 MRI images of 3 brain tumor types (Meningiomas, Gliomas, and Pituitary tumors). We have achieved the highest accuracy of 99\% outperforming the other previous work on the same dataset.},
journal = {Artificial Intelligence in Medicine},
author = {Ismael, Sarah Ali [Abdelaziz and Mohammed, Ammar and Hefny, Hesham},
year = {2020},
keywords = {Artificial neural network, Cancer classification, Convolutional neural network, Deep residual network, Machine learning},
pages = {101779},
}Computer-inspired Quantum Experiments. Krenn, M., Erhard, M., & Zeilinger, A. arXiv:2002.09970 [quant-ph], February, 2020. arXiv: 2002.09970 repo: https://github.com/XuemeiGu/MelvinPython/
Paper abstract bibtex
Paper abstract bibtex The design of new devices and experiments in science and engineering has historically relied on the intuitions of human experts. This credo, however, has changed. In many disciplines, computer-inspired design processes, also known as inverse-design, have augmented the capability of scientists. Here we visit different fields of physics in which computer-inspired designs are applied. We will meet vastly diverse computational approaches based on topological optimization, evolutionary strategies, deep learning, reinforcement learning or automated reasoning. Then we draw our attention specifically on quantum physics. In the quest for designing new quantum experiments, we face two challenges: First, quantum phenomena are unintuitive. Second, the number of possible configurations of quantum experiments explodes combinatorially. To overcome these challenges, physicists began to use algorithms for computer-designed quantum experiments. We focus on the most mature and \textit\practical\ approaches that scientists used to find new complex quantum experiments, which experimentalists subsequently have realized in the laboratories. The underlying idea is a highly-efficient topological search, which allows for scientific interpretability. In that way, some of the computer-designs have led to the discovery of new scientific concepts and ideas – demonstrating how computer algorithm can genuinely contribute to science by providing unexpected inspirations. We discuss several extensions and alternatives based on optimization and machine learning techniques, with the potential of accelerating the discovery of practical computer-inspired experiments or concepts in the future. Finally, we discuss what we can learn from the different approaches in the fields of physics, and raise several fascinating possibilities for future research.
@article{krenn_computer-inspired_2020,
title = {Computer-inspired {Quantum} {Experiments}},
url = {http://arxiv.org/abs/2002.09970},
abstract = {The design of new devices and experiments in science and engineering has historically relied on the intuitions of human experts. This credo, however, has changed. In many disciplines, computer-inspired design processes, also known as inverse-design, have augmented the capability of scientists. Here we visit different fields of physics in which computer-inspired designs are applied. We will meet vastly diverse computational approaches based on topological optimization, evolutionary strategies, deep learning, reinforcement learning or automated reasoning. Then we draw our attention specifically on quantum physics. In the quest for designing new quantum experiments, we face two challenges: First, quantum phenomena are unintuitive. Second, the number of possible configurations of quantum experiments explodes combinatorially. To overcome these challenges, physicists began to use algorithms for computer-designed quantum experiments. We focus on the most mature and {\textbackslash}textit\{practical\} approaches that scientists used to find new complex quantum experiments, which experimentalists subsequently have realized in the laboratories. The underlying idea is a highly-efficient topological search, which allows for scientific interpretability. In that way, some of the computer-designs have led to the discovery of new scientific concepts and ideas -- demonstrating how computer algorithm can genuinely contribute to science by providing unexpected inspirations. We discuss several extensions and alternatives based on optimization and machine learning techniques, with the potential of accelerating the discovery of practical computer-inspired experiments or concepts in the future. Finally, we discuss what we can learn from the different approaches in the fields of physics, and raise several fascinating possibilities for future research.},
urldate = {2020-03-01},
journal = {arXiv:2002.09970 [quant-ph]},
author = {Krenn, Mario and Erhard, Manuel and Zeilinger, Anton},
month = feb,
year = {2020},
note = {arXiv: 2002.09970
repo: https://github.com/XuemeiGu/MelvinPython/},
keywords = {Computer Science - Neural and Evolutionary Computing, machine learning, quantum physics, uses sympy},
}Thermodynamics-based Artificial Neural Networks for constitutive modeling. Masi, F., Stefanou, I., Vannucci, P., & Maffi-Berthier, V. arXiv:2005.12183 [physics, stat], May, 2020. arXiv: 2005.12183
Paper abstract bibtex
Paper abstract bibtex Machine Learning methods and, in particular, Artificial Neural Networks (ANNs) have demonstrated promising capabilities in material constitutive modeling. One of the main drawbacks of such approaches is the lack of a rigorous frame based on the laws of physics. This may render physically inconsistent the predictions of a trained network, which can be even dangerous for real applications. Here we propose a new class of data-driven, physics-based, neural networks for constitutive modeling of strain rate independent processes at the material point level, which we define as Thermodynamics-based Artificial Neural Networks (TANNs). The two basic principles of thermodynamics are encoded in the network's architecture by taking advantage of automatic differentiation to compute the numerical derivatives of a network with respect to its inputs. In this way, derivatives of the free-energy, the dissipation rate and their relation with the stress and internal state variables are hardwired in the network. Consequently, our network does not have to identify the underlying pattern of thermodynamic laws during training, reducing the need of large data-sets. Moreover the training is more efficient and robust, and the predictions more accurate. Finally and more important, the predictions remain thermodynamically consistent, even for unseen data. Based on these features, TANNs are a starting point for data-driven, physics-based constitutive modeling with neural networks. We demonstrate the wide applicability of TANNs for modeling elasto-plastic materials, with strain hardening and strain softening. Detailed comparisons show that the predictions of TANNs outperform those of standard ANNs. TANNs ' architecture is general, enabling applications to materials with different or more complex behavior, without any modification.
@article{masi_thermodynamics-based_2020,
title = {Thermodynamics-based {Artificial} {Neural} {Networks} for constitutive modeling},
url = {http://arxiv.org/abs/2005.12183},
abstract = {Machine Learning methods and, in particular, Artificial Neural Networks (ANNs) have demonstrated promising capabilities in material constitutive modeling. One of the main drawbacks of such approaches is the lack of a rigorous frame based on the laws of physics. This may render physically inconsistent the predictions of a trained network, which can be even dangerous for real applications. Here we propose a new class of data-driven, physics-based, neural networks for constitutive modeling of strain rate independent processes at the material point level, which we define as Thermodynamics-based Artificial Neural Networks (TANNs). The two basic principles of thermodynamics are encoded in the network's architecture by taking advantage of automatic differentiation to compute the numerical derivatives of a network with respect to its inputs. In this way, derivatives of the free-energy, the dissipation rate and their relation with the stress and internal state variables are hardwired in the network. Consequently, our network does not have to identify the underlying pattern of thermodynamic laws during training, reducing the need of large data-sets. Moreover the training is more efficient and robust, and the predictions more accurate. Finally and more important, the predictions remain thermodynamically consistent, even for unseen data. Based on these features, TANNs are a starting point for data-driven, physics-based constitutive modeling with neural networks. We demonstrate the wide applicability of TANNs for modeling elasto-plastic materials, with strain hardening and strain softening. Detailed comparisons show that the predictions of TANNs outperform those of standard ANNs. TANNs ' architecture is general, enabling applications to materials with different or more complex behavior, without any modification.},
urldate = {2020-05-30},
journal = {arXiv:2005.12183 [physics, stat]},
author = {Masi, Filippo and Stefanou, Ioannis and Vannucci, Paolo and Maffi-Berthier, Victor},
month = may,
year = {2020},
note = {arXiv: 2005.12183},
keywords = {Physics - Computational Physics, machine learning},
}Not All Roads Lead to Rome: Pitch Representation and Model Architecture for Automatic Harmonic Analysis. Micchi, G., Gotham, M., & Giraud, M. Transactions of the International Society for Music Information Retrieval, 3(1):42–54, 2020.
doi abstract bibtex
doi abstract bibtex
Automatic harmonic analysis has been an enduring focus of the MIR community, and has enjoyed a particularly vigorous revival of interest in the machine-learning age. We focus here on the specific case of Roman numeral analysis which, by virtue of requiring key/functional information in addition to chords, may be viewed as an acutely challenging use case. We report on three main developments. First, we provide a new meta-corpus bringing together all existing Roman numeral analysis datasets; this offers greater scale and diversity, not only of the music represented, but also of human analytical viewpoints. Second, we examine best practices in the encoding of pitch, time, and harmony for machine learning tasks. The main contribution here is the introduction of full pitch spelling to such a system, an absolute must for the comprehensive study of musical harmony. Third, we devised and tested several neural network architectures and compared their relative accuracy. In the best-performing of these models, convolutional layers gather the local information needed to analyse the chord at a given moment while a recurrent part learns longer-range harmonic progressions. Altogether, our best representation and architecture produce a small but significant improvement on overall accuracy while simultaneously integrating full pitch spelling. This enables the system to retain important information from the musical sources and provide more meaningful predictions for any new input.
@Article{ micchi.ea2020-not,
author = {Micchi, Gianluca and Gotham, Mark and Giraud, Mathieu},
year = {2020},
title = {Not All Roads Lead to Rome: Pitch Representation and
Model Architecture for Automatic Harmonic Analysis},
abstract = {Automatic harmonic analysis has been an enduring focus of
the MIR community, and has enjoyed a particularly vigorous
revival of interest in the machine-learning age. We focus
here on the specific case of Roman numeral analysis which,
by virtue of requiring key/functional information in
addition to chords, may be viewed as an acutely
challenging use case. We report on three main
developments. First, we provide a new meta-corpus bringing
together all existing Roman numeral analysis datasets;
this offers greater scale and diversity, not only of the
music represented, but also of human analytical
viewpoints. Second, we examine best practices in the
encoding of pitch, time, and harmony for machine learning
tasks. The main contribution here is the introduction of
full pitch spelling to such a system, an absolute must for
the comprehensive study of musical harmony. Third, we
devised and tested several neural network architectures
and compared their relative accuracy. In the
best-performing of these models, convolutional layers
gather the local information needed to analyse the chord
at a given moment while a recurrent part learns
longer-range harmonic progressions. Altogether, our best
representation and architecture produce a small but
significant improvement on overall accuracy while
simultaneously integrating full pitch spelling. This
enables the system to retain important information from
the musical sources and provide more meaningful
predictions for any new input.},
doi = {10.5334/tismir.45},
journal = {Transactions of the International Society for Music
Information Retrieval},
keywords = {1,1 key,chords and functional harmony,computational
musicology,corpus,functional harmony,introduction,is
common to a,machine learning,motivation,pitch
encoding,previous work,roman numeral analysis,some sense
of,tonal harmony,very wide},
mendeley-tags= {computational musicology},
number = {1},
pages = {42--54},
volume = {3}
}Pathological test type and chemical detection using deep neural networks: a case study using ELISA and LFA assays. Hoque Tania, M., Kaiser, M. S., Abu-Hassan, K., & Hossain, M. A. Journal of Enterprise Information Management, January, 2020.
Paper doi abstract bibtex
Paper doi abstract bibtex Purpose The gradual increase in geriatric issues and global imbalance of the ratio between patients and healthcare professionals have created a demand for intelligent systems with the least error-prone diagnosis results to be used by less medically trained persons and save clinical time. This paper aims at investigating the development of image-based colourimetric analysis. The purpose of recognising such tests is to support wider users to begin a colourimetric test to be used at homecare settings, telepathology and so on. Design/methodology/approach The concept of an automatic colourimetric assay detection is delivered by utilising two cases. Training deep learning (DL) models on thousands of images of these tests using transfer learning, this paper (1) classifies the type of the assay and (2) classifies the colourimetric results. Findings This paper demonstrated that the assay type can be recognised using DL techniques with 100% accuracy within a fraction of a second. Some of the advantages of the pre-trained model over the calibration-based approach are robustness, readiness and suitability to deploy for similar applications within a shorter period of time. Originality/value To the best of the authors’ knowledge, this is the first attempt to provide colourimetric assay type classification (CATC) using DL. Humans are capable to learn thousands of visual classifications in their life. Object recognition may be a trivial task for humans, due to photometric and geometric variabilities along with the high degree of intra-class variabilities, it can be a challenging task for machines. However, transforming visual knowledge into machines, as proposed, can support non-experts to better manage their health and reduce some of the burdens on experts.
@article{hoque_tania_pathological_2020,
title = {Pathological test type and chemical detection using deep neural networks: a case study using {ELISA} and {LFA} assays},
volume = {ahead-of-print},
issn = {1741-0398},
shorttitle = {Pathological test type and chemical detection using deep neural networks},
url = {https://doi.org/10.1108/JEIM-01-2020-0038},
doi = {10.1108/JEIM-01-2020-0038},
abstract = {Purpose The gradual increase in geriatric issues and global imbalance of the ratio between patients and healthcare professionals have created a demand for intelligent systems with the least error-prone diagnosis results to be used by less medically trained persons and save clinical time. This paper aims at investigating the development of image-based colourimetric analysis. The purpose of recognising such tests is to support wider users to begin a colourimetric test to be used at homecare settings, telepathology and so on. Design/methodology/approach The concept of an automatic colourimetric assay detection is delivered by utilising two cases. Training deep learning (DL) models on thousands of images of these tests using transfer learning, this paper (1) classifies the type of the assay and (2) classifies the colourimetric results. Findings This paper demonstrated that the assay type can be recognised using DL techniques with 100\% accuracy within a fraction of a second. Some of the advantages of the pre-trained model over the calibration-based approach are robustness, readiness and suitability to deploy for similar applications within a shorter period of time. Originality/value To the best of the authors’ knowledge, this is the first attempt to provide colourimetric assay type classification (CATC) using DL. Humans are capable to learn thousands of visual classifications in their life. Object recognition may be a trivial task for humans, due to photometric and geometric variabilities along with the high degree of intra-class variabilities, it can be a challenging task for machines. However, transforming visual knowledge into machines, as proposed, can support non-experts to better manage their health and reduce some of the burdens on experts.},
number = {ahead-of-print},
urldate = {2021-02-21},
journal = {Journal of Enterprise Information Management},
author = {Hoque Tania, Marzia and Kaiser, M. Shamim and Abu-Hassan, Kamal and Hossain, M. A.},
month = jan,
year = {2020},
keywords = {Colourimetric test, Computer vision, Diagnosis, Machine learning, Point-of-care system, Pre-trained model},
}Comparison of Time Series Clustering Algorithms for Machine State Detection. Hennig, M., Grafinger, M., Gerhard, D., Dumss, S., & Rosenberger, P. Procedia CIRP, 93:1352–1357, January, 2020.
Paper doi abstract bibtex
Paper doi abstract bibtex New developments in domains like mathematics and statistical learning and availability of easy-to-use, often freely accessible software tools offer great potential to transform the manufacturing domain and their grasp on the increased manufacturing data repositories sustainably. One of the most exciting developments is in the area of machine learning. Time series clustering could be utilized in machine state detection which can be used in predictive maintenance or online optimization. This paper presents a comparison of freely available time series clustering algorithms, by applying several combinations of different algorithms to a database of public benchmark technical data.
@article{hennig_comparison_2020,
series = {53rd {CIRP} {Conference} on {Manufacturing} {Systems} 2020},
title = {Comparison of {Time} {Series} {Clustering} {Algorithms} for {Machine} {State} {Detection}},
volume = {93},
issn = {2212-8271},
url = {http://www.sciencedirect.com/science/article/pii/S2212827120307149},
doi = {10.1016/j.procir.2020.03.084},
abstract = {New developments in domains like mathematics and statistical learning and availability of easy-to-use, often freely accessible software tools offer great potential to transform the manufacturing domain and their grasp on the increased manufacturing data repositories sustainably. One of the most exciting developments is in the area of machine learning. Time series clustering could be utilized in machine state detection which can be used in predictive maintenance or online optimization. This paper presents a comparison of freely available time series clustering algorithms, by applying several combinations of different algorithms to a database of public benchmark technical data.},
language = {en},
urldate = {2020-09-28},
journal = {Procedia CIRP},
author = {Hennig, Martin and Grafinger, Manfred and Gerhard, Detlef and Dumss, Stefan and Rosenberger, Patrick},
month = jan,
year = {2020},
keywords = {Industry 4.0, Internet of Things, Machine Learning, Predictive Maintenance, Time Series Clustering, Unsupervised Learning},
pages = {1352--1357},
}2019 (14)
Identification of Caveolin-1 Domain Signatures via Graphlet Analysis of Single Molecule Super-Resolution Data. Khater, I. M., Meng, F., Nabi, I. R., & Hamarneh, G. Bioinformatics, 35(18):3468-3475, 2019.
doi bibtex
doi bibtex
@ARTICLE{bioinformatics2019,
AUTHOR = {Ismail M. Khater and Fanrui Meng and Ivan Robert Nabi and
Ghassan Hamarneh},
JOURNAL = {Bioinformatics},
OPTMONTH = {},
OPTNOTE = {},
NUMBER = {18},
PAGES = {3468-3475},
TITLE = {Identification of Caveolin-1 Domain Signatures via
Graphlet Analysis of Single Molecule Super-Resolution Data},
VOLUME = {35},
YEAR = {2019},
OPTABSTRACT = {},
DOI = {10.1093/bioinformatics/btz113},
OPTISBN = {},
OPTISSN = {},
KEYWORDS = {Super Resolution Microscopy,
Single Molecule Localization Microscopy,
Network Modelling and Analysis, Machine Learning, Graphlets,
Graph based, Biomarkers/Biosignatures},
OPTURL = {},
OPTURL-PUBLISHER = {},
PDF = {http://www.cs.sfu.ca/~hamarneh/ecopy/bioinformatics2019.pdf}
}Applications of machine learning in real-life digital health interventions: Review of the literature. Triantafyllidis, A., K. & Tsanas, A. Journal of Medical Internet Research, 21(4):1-9, 2019.
doi abstract bibtex
doi abstract bibtex
Background: Machine learning has attracted considerable research interest toward developing smart digital health interventions. These interventions have the potential to revolutionize health care and lead to substantial outcomes for patients and medical professionals. Objective: Our objective was to review the literature on applications of machine learning in real-life digital health interventions, aiming to improve the understanding of researchers, clinicians, engineers, and policy makers in developing robust and impactful data-driven interventions in the health care domain. Methods: We searched the PubMed and Scopus bibliographic databases with terms related to machine learning, to identify real-life studies of digital health interventions incorporating machine learning algorithms. We grouped those interventions according to their target (ie, target condition), study design, number of enrolled participants, follow-up duration, primary outcome and whether this had been statistically significant, machine learning algorithms used in the intervention, and outcome of the algorithms (eg, prediction). Results: Our literature search identified 8 interventions incorporating machine learning in a real-life research setting, of which 3 (37%) were evaluated in a randomized controlled trial and 5 (63%) in a pilot or experimental single-group study. The interventions targeted depression prediction and management, speech recognition for people with speech disabilities, self-efficacy for weight loss, detection of changes in biopsychosocial condition of patients with multiple morbidity, stress management, treatment of phantom limb pain, smoking cessation, and personalized nutrition based on glycemic response. The average number of enrolled participants in the studies was 71 (range 8-214), and the average follow-up study duration was 69 days (range 3-180). Of the 8 interventions, 6 (75%) showed statistical significance (at the P=.05 level) in health outcomes. Conclusions: This review found that digital health interventions incorporating machine learning algorithms in real-life studies can be useful and effective. Given the low number of studies identified in this review and that they did not follow a rigorous machine learning evaluation methodology, we urge the research community to conduct further studies in intervention settings following evaluation principles and demonstrating the potential of machine learning in clinical practice.
@article{
title = {Applications of machine learning in real-life digital health interventions: Review of the literature},
type = {article},
year = {2019},
keywords = {Artificial intelligence,Data mining,Digital health,Machine learning,Review,Telemedicine},
pages = {1-9},
volume = {21},
id = {4addf79b-282a-34d9-9174-7f7a07d3686d},
created = {2020-04-30T12:14:51.821Z},
file_attached = {false},
profile_id = {63c14ecf-8d9b-3014-983a-e77627cb99e3},
group_id = {5c252342-c3f4-3788-aa2e-3a2bba078fe3},
last_modified = {2020-04-30T12:14:51.821Z},
read = {false},
starred = {false},
authored = {false},
confirmed = {true},
hidden = {false},
private_publication = {false},
abstract = {Background: Machine learning has attracted considerable research interest toward developing smart digital health interventions. These interventions have the potential to revolutionize health care and lead to substantial outcomes for patients and medical professionals. Objective: Our objective was to review the literature on applications of machine learning in real-life digital health interventions, aiming to improve the understanding of researchers, clinicians, engineers, and policy makers in developing robust and impactful data-driven interventions in the health care domain. Methods: We searched the PubMed and Scopus bibliographic databases with terms related to machine learning, to identify real-life studies of digital health interventions incorporating machine learning algorithms. We grouped those interventions according to their target (ie, target condition), study design, number of enrolled participants, follow-up duration, primary outcome and whether this had been statistically significant, machine learning algorithms used in the intervention, and outcome of the algorithms (eg, prediction). Results: Our literature search identified 8 interventions incorporating machine learning in a real-life research setting, of which 3 (37%) were evaluated in a randomized controlled trial and 5 (63%) in a pilot or experimental single-group study. The interventions targeted depression prediction and management, speech recognition for people with speech disabilities, self-efficacy for weight loss, detection of changes in biopsychosocial condition of patients with multiple morbidity, stress management, treatment of phantom limb pain, smoking cessation, and personalized nutrition based on glycemic response. The average number of enrolled participants in the studies was 71 (range 8-214), and the average follow-up study duration was 69 days (range 3-180). Of the 8 interventions, 6 (75%) showed statistical significance (at the P=.05 level) in health outcomes. Conclusions: This review found that digital health interventions incorporating machine learning algorithms in real-life studies can be useful and effective. Given the low number of studies identified in this review and that they did not follow a rigorous machine learning evaluation methodology, we urge the research community to conduct further studies in intervention settings following evaluation principles and demonstrating the potential of machine learning in clinical practice.},
bibtype = {article},
author = {Triantafyllidis, Andreas K. and Tsanas, Athanasios},
doi = {10.2196/12286},
journal = {Journal of Medical Internet Research},
number = {4}
}Next Generation Mapping: Combining Deep Learning, Cloud Computing, and Big Remote Sensing Data. Parente, L., Taquary, E., Silva, A. P., Souza, C., & Ferreira, L. Remote Sensing, 11(23):2881, January, 2019.
Paper doi abstract bibtex
Paper doi abstract bibtex The rapid growth of satellites orbiting the planet is generating massive amounts of data for Earth science applications. Concurrently, state-of-the-art deep-learning-based algorithms and cloud computing infrastructure have become available with a great potential to revolutionize the image processing of satellite remote sensing. Within this context, this study evaluated, based on thousands of PlanetScope images obtained over a 12-month period, the performance of three machine learning approaches (random forest, long short-term memory-LSTM, and U-Net). We applied these approaches to mapped pasturelands in a Central Brazil region. The deep learning algorithms were implemented using TensorFlow, while the random forest utilized the Google Earth Engine platform. The accuracy assessment presented F1 scores for U-Net, LSTM, and random forest of, respectively, 96.94%, 98.83%, and 95.53% in the validation data, and 94.06%, 87.97%, and 82.57% in the test data, indicating a better classification efficiency using the deep learning approaches. Although the use of deep learning algorithms depends on a high investment in calibration samples and the generalization of these methods requires further investigations, our results suggest that the neural network architectures developed in this study can be used to map large geographic regions that consider a wide variety of satellite data (e.g., PlanetScope, Sentinel-2, Landsat-8).
@article{parente_next_2019,
title = {Next {Generation} {Mapping}: {Combining} {Deep} {Learning}, {Cloud} {Computing}, and {Big} {Remote} {Sensing} {Data}},
volume = {11},
copyright = {http://creativecommons.org/licenses/by/3.0/},
shorttitle = {Next {Generation} {Mapping}},
url = {https://www.mdpi.com/2072-4292/11/23/2881},
doi = {10.3390/rs11232881},
abstract = {The rapid growth of satellites orbiting the planet is generating massive amounts of data for Earth science applications. Concurrently, state-of-the-art deep-learning-based algorithms and cloud computing infrastructure have become available with a great potential to revolutionize the image processing of satellite remote sensing. Within this context, this study evaluated, based on thousands of PlanetScope images obtained over a 12-month period, the performance of three machine learning approaches (random forest, long short-term memory-LSTM, and U-Net). We applied these approaches to mapped pasturelands in a Central Brazil region. The deep learning algorithms were implemented using TensorFlow, while the random forest utilized the Google Earth Engine platform. The accuracy assessment presented F1 scores for U-Net, LSTM, and random forest of, respectively, 96.94\%, 98.83\%, and 95.53\% in the validation data, and 94.06\%, 87.97\%, and 82.57\% in the test data, indicating a better classification efficiency using the deep learning approaches. Although the use of deep learning algorithms depends on a high investment in calibration samples and the generalization of these methods requires further investigations, our results suggest that the neural network architectures developed in this study can be used to map large geographic regions that consider a wide variety of satellite data (e.g., PlanetScope, Sentinel-2, Landsat-8).},
language = {en},
number = {23},
urldate = {2019-12-05},
journal = {Remote Sensing},
author = {Parente, Leandro and Taquary, Evandro and Silva, Ana Paula and Souza, Carlos and Ferreira, Laerte},
month = jan,
year = {2019},
keywords = {LSTM, LULC classification, PlanetScope, U-Net, deep learning, machine learning, random forest},
pages = {2881}
}Deep Learning at Scale. Viviani, P., Drocco, M., Baccega, D., Colonnelli, I., & Aldinucci, M. In Proc. of 27th Euromicro Intl. Conference on Parallel Distributed and network-based Processing (PDP), pages 124–131, Pavia, Italy, 2019. IEEE.
Paper doi abstract bibtex
Paper doi abstract bibtex This work presents a novel approach to distributed training of deep neural networks (DNNs) that aims to overcome the issues related to mainstream approaches to data parallel training. Established techniques for data parallel training are discussed from both a parallel computing and deep learning perspective, then a different approach is presented that is meant to allow DNN training to scale while retaining good convergence properties. Moreover, an experimental implementation is presented as well as some preliminary results.
@inproceedings{19:deeplearn:pdp,
title = {Deep Learning at Scale},
author = {Paolo Viviani and Maurizio Drocco and Daniele Baccega and Iacopo Colonnelli and Marco Aldinucci},
year = 2019,
booktitle = {Proc. of 27th Euromicro Intl. Conference on Parallel Distributed and network-based Processing (PDP)},
publisher = {IEEE},
address = {Pavia, Italy},
pages = {124--131},
doi = {10.1109/EMPDP.2019.8671552},
url = {https://iris.unito.it/retrieve/handle/2318/1695211/487778/19_deeplearning_PDP.pdf},
abstract = {This work presents a novel approach to distributed training of deep neural networks (DNNs) that aims to overcome the issues related to mainstream approaches to data parallel training. Established techniques for data parallel training are discussed from both a parallel computing and deep learning perspective, then a different approach is presented that is meant to allow DNN training to scale while retaining good convergence properties. Moreover, an experimental implementation is presented as well as some preliminary results.},
date-modified = {2019-03-22 22:49:35 +0100},
keywords = {deep learning, distributed computing, machine learning, large scale, C++},
bdsk-url-1 = {https://iris.unito.it/retrieve/handle/2318/1695211/487778/19_deeplearning_PDP.pdf}
}Generalizable Feature Learning in the Presence of Data Bias and Domain Class Imbalance with Application to Skin Lesion Classification. Yoon, C., Hamarneh, G., & Abugharbieh, R. In Lecture Notes in Computer Science, Medical Image Computing and Computer-Assisted Intervention (MICCAI), volume 11767, pages 365-373, 2019.
doi bibtex
doi bibtex
@INPROCEEDINGS{miccai2019d,
OPTADDRESS = {},
AUTHOR = {Chris Yoon and Ghassan Hamarneh and Rafeef Abugharbieh},
BOOKTITLE = {Lecture Notes in Computer Science, Medical Image Computing and Computer-Assisted Intervention (MICCAI)},
OPTEDITOR = {},
OPTMONTH = {},
OPTNOTE = {},
OPTNUMBER = {},
OPTORGANIZATION = {},
PAGES = {365-373},
OPTPUBLISHER = {},
OPTSERIES = {},
TITLE = {Generalizable Feature Learning in the Presence of Data
Bias and Domain Class Imbalance with Application to Skin Lesion
Classification},
VOLUME = {11767},
YEAR = {2019},
OPTABSTRACT = {},
DOI = {0.1007/978-3-030-32251-9_40},
OPTISBN = {},
OPTISSN = {},
KEYWORDS = {Classification, Dermatology, Machine Learning,
Deep Learning},
OPTURL = {},
OPTURL-PUBLISHER = {},
PDF = {http://www.cs.sfu.ca/~hamarneh/ecopy/miccai2019d.pdf}
}Aggregated outputs by linear models: An application on marine litter beaching prediction. Hernández-González, J., Inza, I., Granado, I., Basurko, O., C., Fernandes, J., A., & Lozano, J., A. Information Sciences, 481:381-393, 2019.
abstract bibtex
abstract bibtex
In regression, a predictive model which is able to anticipate the output of a new case is learnt from a set of previous examples. The output or response value of these examples used for model training is known. When learning with aggregated outputs, the examples available for model training are individually unlabeled. Collectively, the aggregated outputs of different subsets of training examples are provided. In this paper, we propose an iterative methodology to learn linear models from this type of data. In spite of being simple, its competitive performance is shown in comparison with a straightforward solution and state-of-the-art techniques. A real world problem is also illustrated which naturally fits the aggregated outputs framework: the estimation of marine litter beaching along the south-east coastline of the Bay of Biscay.
@article{
title = {Aggregated outputs by linear models: An application on marine litter beaching prediction},
type = {article},
year = {2019},
keywords = {Aggregated outputs,Expectation–Maximization,Linear models,Machine learning,Marine litter beaching,Regression},
pages = {381-393},
volume = {481},
id = {cc28644d-9826-368f-93dc-a69861cfed15},
created = {2021-11-12T08:30:57.939Z},
file_attached = {false},
profile_id = {789246de-927b-32cc-ae4f-1b7e2b31674c},
group_id = {e3c82d43-35db-3bbb-b28a-0fd521d70498},
last_modified = {2021-11-12T08:30:57.939Z},
read = {false},
starred = {false},
authored = {false},
confirmed = {false},
hidden = {false},
source_type = {article},
private_publication = {false},
abstract = {In regression, a predictive model which is able to anticipate the output of a new case is learnt from a set of previous examples. The output or response value of these examples used for model training is known. When learning with aggregated outputs, the examples available for model training are individually unlabeled. Collectively, the aggregated outputs of different subsets of training examples are provided. In this paper, we propose an iterative methodology to learn linear models from this type of data. In spite of being simple, its competitive performance is shown in comparison with a straightforward solution and state-of-the-art techniques. A real world problem is also illustrated which naturally fits the aggregated outputs framework: the estimation of marine litter beaching along the south-east coastline of the Bay of Biscay.},
bibtype = {article},
author = {Hernández-González, Jerónimo and Inza, Iñaki and Granado, Igor and Basurko, Oihane C and Fernandes, Jose A and Lozano, Jose A},
journal = {Information Sciences}
}Online Machine Learning Audiometry. Barbour, D. L., Howard, R. T., Song, X. D., Metzger, N., Sukesan, K. A., DiLorenzo, J. C., Snyder, B. R. D., Chen, J. Y., Degen, E. A., Buchbinder, J. M., & Heisey, K. L. Ear and Hearing, 40(4):918–926, August, 2019.
Paper doi abstract bibtex
Paper doi abstract bibtex OBJECTIVES: A confluence of recent developments in cloud computing, real-time web audio and machine learning psychometric function estimation has made wide dissemination of sophisticated turn-key audiometric assessments possible. The authors have combined these capabilities into an online (i.e., web-based) pure-tone audiogram estimator intended to empower researchers and clinicians with advanced hearing tests without the need for custom programming or special hardware. The objective of this study was to assess the accuracy and reliability of this new online machine learning audiogram method relative to a commonly used hearing threshold estimation technique also implemented online for the first time in the same platform. DESIGN: The authors performed air conduction pure-tone audiometry on 21 participants between the ages of 19 and 79 years (mean 41, SD 21) exhibiting a wide range of hearing abilities. For each ear, two repetitions of online machine learning audiogram estimation and two repetitions of online modified Hughson-Westlake ascending-descending audiogram estimation were acquired by an audiologist using the online software tools. The estimated hearing thresholds of these two techniques were compared at standard audiogram frequencies (i.e., 0.25, 0.5, 1, 2, 4, 8 kHz). RESULTS: The two threshold estimation methods delivered very similar threshold estimates at standard audiogram frequencies. Specifically, the mean absolute difference between threshold estimates was 3.24 ± 5.15 dB. The mean absolute differences between repeated measurements of the online machine learning procedure and between repeated measurements of the Hughson-Westlake procedure were 2.85 ± 6.57 dB and 1.88 ± 3.56 dB, respectively. The machine learning method generated estimates of both threshold and spread (i.e., the inverse of psychometric slope) continuously across the entire frequency range tested from fewer samples on average than the modified Hughson-Westlake procedure required to estimate six discrete thresholds. CONCLUSIONS: Online machine learning audiogram estimation in its current form provides all the information of conventional threshold audiometry with similar accuracy and reliability in less time. More importantly, however, this method provides additional audiogram details not provided by other methods. This standardized platform can be readily extended to bone conduction, masking, spectrotemporal modulation, speech perception, etc., unifying audiometric testing into a single comprehensive procedure efficient enough to become part of the standard audiologic workup.
@article{barbour_d_l_online_2019,
title = {Online {Machine} {Learning} {Audiometry}},
volume = {40},
issn = {1538-4667},
url = {https://journals.lww.com/ear-hearing/Abstract/2019/07000/Online_Machine_Learning_Audiometry.14.aspx},
doi = {10.1097/AUD.0000000000000669},
abstract = {OBJECTIVES: A confluence of recent developments in cloud computing, real-time web audio and machine learning psychometric function estimation has made wide dissemination of sophisticated turn-key audiometric assessments possible. The authors have combined these capabilities into an online (i.e., web-based) pure-tone audiogram estimator intended to empower researchers and clinicians with advanced hearing tests without the need for custom programming or special hardware. The objective of this study was to assess the accuracy and reliability of this new online machine learning audiogram method relative to a commonly used hearing threshold estimation technique also implemented online for the first time in the same platform.
DESIGN: The authors performed air conduction pure-tone audiometry on 21 participants between the ages of 19 and 79 years (mean 41, SD 21) exhibiting a wide range of hearing abilities. For each ear, two repetitions of online machine learning audiogram estimation and two repetitions of online modified Hughson-Westlake ascending-descending audiogram estimation were acquired by an audiologist using the online software tools. The estimated hearing thresholds of these two techniques were compared at standard audiogram frequencies (i.e., 0.25, 0.5, 1, 2, 4, 8 kHz).
RESULTS: The two threshold estimation methods delivered very similar threshold estimates at standard audiogram frequencies. Specifically, the mean absolute difference between threshold estimates was 3.24 ± 5.15 dB. The mean absolute differences between repeated measurements of the online machine learning procedure and between repeated measurements of the Hughson-Westlake procedure were 2.85 ± 6.57 dB and 1.88 ± 3.56 dB, respectively. The machine learning method generated estimates of both threshold and spread (i.e., the inverse of psychometric slope) continuously across the entire frequency range tested from fewer samples on average than the modified Hughson-Westlake procedure required to estimate six discrete thresholds.
CONCLUSIONS: Online machine learning audiogram estimation in its current form provides all the information of conventional threshold audiometry with similar accuracy and reliability in less time. More importantly, however, this method provides additional audiogram details not provided by other methods. This standardized platform can be readily extended to bone conduction, masking, spectrotemporal modulation, speech perception, etc., unifying audiometric testing into a single comprehensive procedure efficient enough to become part of the standard audiologic workup.},
language = {eng},
number = {4},
journal = {Ear and Hearing},
author = {{Barbour, D. L.} and Howard, Rebecca T. and Song, Xinyu D. and Metzger, Nikki and Sukesan, Kiron A. and DiLorenzo, James C. and Snyder, Braham R. D. and Chen, Jeff Y. and Degen, Eleanor A. and Buchbinder, Jenna M. and Heisey, Katherine L.},
month = aug,
year = {2019},
pmid = {30358656},
pmcid = {PMC6476703},
keywords = {Adult, Aged, Audiometry, Pure-Tone, Female, Hearing Loss, Humans, Internet, Machine Learning, Male, Middle Aged, Reproducibility of Results, Severity of Illness Index, Young Adult},
pages = {918--926},
}From open access to perpetual access: archiving web-published scholarship. Bailey, J. & Praetzellis, M. June, 2019.
Paper bibtex
Paper bibtex @misc{bailey_open_2019,
address = {Zagreb},
title = {From open access to perpetual access: archiving web-published scholarship},
url = {https://digital.library.unt.edu/ark:/67531/metadc1609026/},
author = {Bailey, Jefferson and Praetzellis, Maria},
month = jun,
year = {2019},
keywords = {FatCat, IIPC WAC2019, Internet Archive, Zagreb, machine learning, web archiving},
}Machine learning discovery of longitudinal patterns of depression and suicidal ideation. Gong, J., Simon, G. E., & Liu, S. PLOS ONE, 14(9):e0222665, September, 2019.
Paper doi bibtex
Paper doi bibtex @article{gong_machine_2019,
title = {Machine learning discovery of longitudinal patterns of depression and suicidal ideation},
volume = {14},
issn = {1932-6203},
url = {https://dx.plos.org/10.1371/journal.pone.0222665},
doi = {10.1371/journal.pone.0222665},
language = {en},
number = {9},
urldate = {2023-03-13},
journal = {PLOS ONE},
author = {Gong, Jue and Simon, Gregory E. and Liu, Shan},
editor = {De Luca, Vincenzo},
month = sep,
year = {2019},
pmcid = {PMC6754154},
pmid = {31539408},
keywords = {AE, autoencoder, ML, Machine Learning, NN, Neural Networks, PHQ-8, PHQ-9, Python, Theano, scikit-learn},
pages = {e0222665},
}Machine Learning Prediction of Defect Types for Electroluminescence Images of Photovoltaic Panels. Mantel, C., Villebro, F., Benatto, G., Parikh, H., Wendlandt, S., Hossain, K., Poulsen, P., Spataru, S., Sera, D., & Forchhammer, S. In Proceedings of SPIE, volume 11139, of Proceedings of SPIE, the International Society for Optical Engineering, 2019. SPIE - International Society for Optical Engineering. 14th International Conference on Solid State Lighting and LED-based Illumination Systems
: SPIE Optical Engineering + Applications ; Conference date: 09-08-2015 Through 13-08-2015
doi abstract bibtex
: SPIE Optical Engineering + Applications ; Conference date: 09-08-2015 Through 13-08-2015
doi abstract bibtex
Despite recent technological advances for Photovoltaic panels maintenance (Electroluminescence imaging, drone inspection), only few large-scale studies achieve identification of the precise category of defects or faults. In this work, Electroluminescence imaged modules are automatically split into cells using projections on the x and y axes to detect cell boundaries. Regions containing potential defects or faults are then detected using Hough transform combined with mathematical morphology. Care is taken to remove most of the bus bars or cell boundaries. Afterwards, 25 features are computed, focusing on both the geometry of the regions (e.g. area, perimeter, circularity) and the statistical characteristics of their pixel values (e.g. median, standard deviation, skewness). Finally, features are mapped to the ground truth labels with Support Vector Machine (RBF kernel) and Random Forest algorithms, coupled with undersampling and SMOTE oversampling, with a stratified 5- folds approach for cross validation. A dataset of 982 Electroluminescence images of installed multi-crystalline photovoltaic modules was acquired in outdoor conditions (evening) with a CMOS sensor. After automatic blur detection, 753 images or 47244 cells remain to evaluate faults. All images were evaluated by experts in PV fault detection that labelled: Finger failures, and three types of cracks based on their respective severity levels (A, B and C). Our results based on 6 data series, yield using Support Vector Machine an accuracy of 0.997 and a recall of 0.274. Improving the region detection process will most likely allow improving the performance.
@inproceedings{7eceed1439874b82a5e48897fae0e5f0,
title = "Machine Learning Prediction of Defect Types for Electroluminescence Images of Photovoltaic Panels",
abstract = "Despite recent technological advances for Photovoltaic panels maintenance (Electroluminescence imaging, drone inspection), only few large-scale studies achieve identification of the precise category of defects or faults. In this work, Electroluminescence imaged modules are automatically split into cells using projections on the x and y axes to detect cell boundaries. Regions containing potential defects or faults are then detected using Hough transform combined with mathematical morphology. Care is taken to remove most of the bus bars or cell boundaries. Afterwards, 25 features are computed, focusing on both the geometry of the regions (e.g. area, perimeter, circularity) and the statistical characteristics of their pixel values (e.g. median, standard deviation, skewness). Finally, features are mapped to the ground truth labels with Support Vector Machine (RBF kernel) and Random Forest algorithms, coupled with undersampling and SMOTE oversampling, with a stratified 5- folds approach for cross validation. A dataset of 982 Electroluminescence images of installed multi-crystalline photovoltaic modules was acquired in outdoor conditions (evening) with a CMOS sensor. After automatic blur detection, 753 images or 47244 cells remain to evaluate faults. All images were evaluated by experts in PV fault detection that labelled: Finger failures, and three types of cracks based on their respective severity levels (A, B and C). Our results based on 6 data series, yield using Support Vector Machine an accuracy of 0.997 and a recall of 0.274. Improving the region detection process will most likely allow improving the performance.",
keywords = "Machine learning, Solar panel, Defect detection, Fault detection, Electroluminescence imaging",
author = "Claire Mantel and Frederik Villebro and Benatto, {Gisele Alves dos Reis} and Parikh, {Harsh Rajesh} and Stefan Wendlandt and Kabir Hossain and Poulsen, {Peter Behrensdorff} and Sergiu Spataru and Dezso Sera and S{\o}ren Forchhammer",
year = "2019",
doi = "10.1117/12.2528440",
language = "English",
volume = "11139",
series = "Proceedings of SPIE, the International Society for Optical Engineering",
publisher = "SPIE - International Society for Optical Engineering",
booktitle = "Proceedings of SPIE",
note = "14th International Conference on Solid State Lighting and LED-based Illumination Systems<br/> : SPIE Optical Engineering + Applications ; Conference date: 09-08-2015 Through 13-08-2015",
}Data-Driven Chemical Reaction Classification, Fingerprinting and Clustering using Attention-Based Neural Networks. Philippe Schwaller, Daniel Probst, Alain C. Vaucher, Vishnu H Nair, Teodoro Laino, & Jean-Louis Reymond ChemRxiv, December, 2019.
Paper doi abstract bibtex
Paper doi abstract bibtex Organic reactions are usually assigned to classes grouping reactions with similar reagents and mechanisms. The classification process is a tedious task, requiring first an accurate mapping of the reaction (atom mapping) followed by the identification of the corresponding reaction class template. In this work, we present two transformer-based models that infer reaction classes from the SMILES representation of chemical reactions. Our best model reaches a classification accuracy of 98.2%. We study the incorrect predictions of the models and show that they reveal different biases and mistakes in the underlying data set. Using the embeddings of our classification model, we introduce reaction fingerprints that do not require knowing the reaction center or distinguishing between reactants and reagents. This conversion from chemical reactions to feature vectors enables efficient clustering and similarity search in the reaction space. We compare the reaction clustering for combinations of self-supervised, supervised, and molecular shingle-based reaction representations.
@article{philippe_schwaller_data-driven_2019,
title = {Data-{Driven} {Chemical} {Reaction} {Classification}, {Fingerprinting} and {Clustering} using {Attention}-{Based} {Neural} {Networks}},
url = {https://chemrxiv.org/articles/Data-Driven_Chemical_Reaction_Classification_with_Attention-Based_Neural_Networks/9897365},
doi = {10.26434/chemrxiv.9897365.v2},
abstract = {Organic reactions are usually assigned to classes grouping reactions with similar reagents and mechanisms. The classification process is a tedious task, requiring first an accurate mapping of the reaction (atom mapping) followed by the identification of the corresponding reaction class template. In this work, we present two transformer-based models that infer reaction classes from the SMILES representation of chemical reactions. Our best model reaches a classification accuracy of 98.2\%. We study the incorrect predictions of the models and show that they reveal different biases and mistakes in the underlying data set. Using the embeddings of our classification model, we introduce reaction fingerprints that do not require knowing the reaction center or distinguishing between reactants and reagents. This conversion from chemical reactions to feature vectors enables efficient clustering and similarity search in the reaction space. We compare the reaction clustering for combinations of self-supervised, supervised, and molecular shingle-based reaction representations.},
journal = {ChemRxiv},
author = {{Philippe Schwaller} and {Daniel Probst} and {Alain C. Vaucher} and {Vishnu H Nair} and {Teodoro Laino} and {Jean-Louis Reymond}},
month = dec,
year = {2019},
keywords = {BERT, Chemical Reactions, Clustering analysis, Fingerprints, SMILES, SMILES string representation, SMILES-Encoded Molecular Structures, classification, deep learning, machine learning, organic chemistry, organic synthesis, transformer},
}Multi-dimensional Evaluation of Temporal Neural Networks on Solar Irradiance Forecasting. Song, Z. & Brown, L. E. In 2019 IEEE Power Energy Society Innovative Smart Grid Technologies Conference (ISGT)-Asia, Chengdu, China, May, 2019.
doi bibtex 1 download
doi bibtex 1 download
@inproceedings{Song.Brown.ISGT-Asia.2019,
author = {Zhenyu Song and Laura E. Brown},
author_short = {Song, Z. and Brown, L. E.},
title = {Multi-dimensional Evaluation of Temporal Neural Networks on Solar Irradiance Forecasting},
booktitle = {2019 IEEE Power Energy Society Innovative Smart Grid Technologies Conference (ISGT)-Asia},
year = {2019},
address = {Chengdu, China},
month = {May},
doi = {10.1109/ISGT-Asia.2019.8881784},
text = {ISGT Asia 2019},
type = {{Conference Paper}},
bibtype = {inproceedings},
keywords = {{Power Systems}, {Machine Learning}}
}SnapCode - A Snapshot Based Approach to Code Stylometry. Sarnot, S. A. P., Rinke, S., Raimalwalla, R., Joshi, R., Khengare, R., & Goel, P. In 2019 International Conference on Information Technology (ICIT), pages 337–341, December, 2019.
doi abstract bibtex
doi abstract bibtex
Artificial neural networks have seen significant advancements in recent times with the growing popularity of deep learning. Deep learning allows us to learn representations that are otherwise difficult to extract and helps in better classification tasks. Images, videos and speech processing are the major areas where deep learning is applied. Our work is related to the application of deep learning to source codes. Previous works in this domain have failed to easily capture structural and behavioral aspects of the code. Thereby relying on manual feature engineering for applications like author identification, code quality analysis, cyber-attack investigation, malware recognition and plagiarism detection. We propose a novel approach to capture these feature representations by processing snapshots of code instead of processing source code token by token. We, therefore, propose SnapCode, a snapshot-based approach to extract deep convolutional features from text which would otherwise be impossible using currently known approaches. SnapCode uses a deep convolutional neural network coupled with transfer learning to learn the structural representation of the source code. We show that simple networks fail to learn these features and deep network coupled with transfer learning gives us the best results. SnapCode can capture behavioral aspects of source code as we employ it to the task of author detection, also known as "code stylometry". We choose author detection to validate our approach as it requires most number of manual and complicated features. Although source code is simply text, we aim to process text data in a way similar to humans and show that we could learn meaningful representations.
@inproceedings{sarnot_snapcode_2019,
title = {{SnapCode} - {A} {Snapshot} {Based} {Approach} to {Code} {Stylometry}},
doi = {10.1109/ICIT48102.2019.00066},
abstract = {Artificial neural networks have seen significant advancements in recent times with the growing popularity of deep learning. Deep learning allows us to learn representations that are otherwise difficult to extract and helps in better classification tasks. Images, videos and speech processing are the major areas where deep learning is applied. Our work is related to the application of deep learning to source codes. Previous works in this domain have failed to easily capture structural and behavioral aspects of the code. Thereby relying on manual feature engineering for applications like author identification, code quality analysis, cyber-attack investigation, malware recognition and plagiarism detection. We propose a novel approach to capture these feature representations by processing snapshots of code instead of processing source code token by token. We, therefore, propose SnapCode, a snapshot-based approach to extract deep convolutional features from text which would otherwise be impossible using currently known approaches. SnapCode uses a deep convolutional neural network coupled with transfer learning to learn the structural representation of the source code. We show that simple networks fail to learn these features and deep network coupled with transfer learning gives us the best results. SnapCode can capture behavioral aspects of source code as we employ it to the task of author detection, also known as "code stylometry". We choose author detection to validate our approach as it requires most number of manual and complicated features. Although source code is simply text, we aim to process text data in a way similar to humans and show that we could learn meaningful representations.},
booktitle = {2019 {International} {Conference} on {Information} {Technology} ({ICIT})},
author = {Sarnot, Saloni Alias Puja and Rinke, Sanjana and Raimalwalla, Rayomand and Joshi, Raviraj and Khengare, Rahul and Goel, Purvi},
month = dec,
year = {2019},
keywords = {Code Stylometry, Computer languages, Convolutional Neural Network, Convolutional neural networks, Feature extraction, Image Processing, Machine learning, Manuals, Syntactics, Task analysis, Transfer learning},
pages = {337--341},
}Learning Everywhere: Pervasive Machine Learning for Effective High-Performance Computation. Fox, G., Glazier, J. A., Kadupitiya, J., Jadhao, V., Kim, M., Qiu, J., Sluka, J. P., Somogyi, E., Marathe, M., Adiga, A., Chen, J., Beckstein, O., & Jha, S. In 2019 IEEE International Parallel and Distributed Processing Symposium Workshops (IPDPSW), pages 422–429, May, 2019. ISSN: null
doi abstract bibtex
doi abstract bibtex
The convergence of HPC and data intensive methodologies provide a promising approach to major performance improvements. This paper provides a general description of the interaction between traditional HPC and ML approaches and motivates the "Learning Everywhere" paradigm for HPC. We introduce the concept of "effective performance" that one can achieve by combining learning methodologies with simulation based approaches, and distinguish between traditional performance as measured by benchmark scores. To support the promise of integrating HPC and learning methods, this paper examines specific examples and opportunities across a series of domains. It concludes with a series of open software systems, methods and infrastructure challenges that the Learning Everywhere paradigm presents.
@inproceedings{fox_learning_2019,
title = {Learning {Everywhere}: {Pervasive} {Machine} {Learning} for {Effective} {High}-{Performance} {Computation}},
shorttitle = {Learning {Everywhere}},
doi = {10.1109/IPDPSW.2019.00081},
abstract = {The convergence of HPC and data intensive methodologies provide a promising approach to major performance improvements. This paper provides a general description of the interaction between traditional HPC and ML approaches and motivates the "Learning Everywhere" paradigm for HPC. We introduce the concept of "effective performance" that one can achieve by combining learning methodologies with simulation based approaches, and distinguish between traditional performance as measured by benchmark scores. To support the promise of integrating HPC and learning methods, this paper examines specific examples and opportunities across a series of domains. It concludes with a series of open software systems, methods and infrastructure challenges that the Learning Everywhere paradigm presents.},
booktitle = {2019 {IEEE} {International} {Parallel} and {Distributed} {Processing} {Symposium} {Workshops} ({IPDPSW})},
author = {Fox, Geoffrey and Glazier, James A. and Kadupitiya, J.C.S. and Jadhao, Vikram and Kim, Minje and Qiu, Judy and Sluka, James P. and Somogyi, Endre and Marathe, Madhav and Adiga, Abhijin and Chen, Jiangzhuo and Beckstein, Oliver and Jha, Shantenu},
month = may,
year = {2019},
note = {ISSN: null},
keywords = {Biological system modeling, Computational modeling, Data models, Effective Performance, Forecasting, ML approaches, Machine learning, Machine learning driven HPC, Mathematical model, Predictive models, data intensive methodologies, general description, high-performance computation, learning (artificial intelligence), learning everywhere paradigm, learning methodologies, learning methods, open software systems, parallel processing, performance improvements, pervasive machine learning, simulation based approaches, traditional HPC, traditional performance, ubiquitous computing},
pages = {422--429},
}2018 (10)
Learning Fast and Slow: A Unified Batch/Stream Framework. Montiel, J., Bifet, A., Losing, V., Read, J., & Abdessalem, T. In 2018 IEEE International Conference on Big Data (Big Data), pages 1065–1072, December, 2018.
doi abstract bibtex
doi abstract bibtex
Data ubiquity highlights the need of efficient and adaptable data-driven solutions. In this paper, we present FAST AND SLOW LEARNING (FSL), a novel unified framework that sheds light on the symbiosis between batch and stream learning. FSL works by employing Fast (stream) and Slow (batch) Learners, emulating the mechanisms used by humans to make decisions. We showcase the applicability of FSL on the task of classification by introducing the FAST AND SLOW CLASSIFIER (FSC). A Fast Learner provides predictions on the spot, continuously updating its model and adapting to changes in the data. On the other hand, the Slow Learner provides predictions considering a wider spectrum of seen data, requiring more time and data to create complex models. Once that enough data has been collected, FSC trains the Slow Learner and starts tracking the performance of both learners. A drift detection mechanism triggers the creation of new Slow models when the current Slow model becomes obsolete. FSC selects between Fast and Slow Learners according to their performance on new incoming data. Test results on real and synthetic data show that FSC effectively drives the positive interaction of stream and batch models for learning from evolving data streams.
@inproceedings{montiel_learning_2018,
title = {Learning {Fast} and {Slow}: {A} {Unified} {Batch}/{Stream} {Framework}},
shorttitle = {Learning {Fast} and {Slow}},
doi = {10.1109/BigData.2018.8622222},
abstract = {Data ubiquity highlights the need of efficient and adaptable data-driven solutions. In this paper, we present FAST AND SLOW LEARNING (FSL), a novel unified framework that sheds light on the symbiosis between batch and stream learning. FSL works by employing Fast (stream) and Slow (batch) Learners, emulating the mechanisms used by humans to make decisions. We showcase the applicability of FSL on the task of classification by introducing the FAST AND SLOW CLASSIFIER (FSC). A Fast Learner provides predictions on the spot, continuously updating its model and adapting to changes in the data. On the other hand, the Slow Learner provides predictions considering a wider spectrum of seen data, requiring more time and data to create complex models. Once that enough data has been collected, FSC trains the Slow Learner and starts tracking the performance of both learners. A drift detection mechanism triggers the creation of new Slow models when the current Slow model becomes obsolete. FSC selects between Fast and Slow Learners according to their performance on new incoming data. Test results on real and synthetic data show that FSC effectively drives the positive interaction of stream and batch models for learning from evolving data streams.},
booktitle = {2018 {IEEE} {International} {Conference} on {Big} {Data} ({Big} {Data})},
author = {Montiel, Jacob and Bifet, Albert and Losing, Viktor and Read, Jesse and Abdessalem, Talel},
month = dec,
year = {2018},
keywords = {Adaptation models, Batch Learning, Classification, Concept Drift, Data models, Machine Learning, Machine learning, Power capacitors, Predictive models, Stream Learning, Task analysis, Training},
pages = {1065--1072},
}Multiobjective Optimization for Stiffness and Position Control in a Soft Robot Arm Module. Ansari, Y., Manti, M., Falotico, E., Cianchetti, M., & Laschi, C. IEEE Robotics and Automation Letters, 3(1):108–115, January, 2018.
doi abstract bibtex
doi abstract bibtex
The central concept of this letter is to develop an assistive manipulator that can automate the bathing task for elderly citizens. We propose to exploit principles of soft robotic technologies to design and control a compliant system to ensure safe human–robot interaction, a primary requirement for the task. The overall system is intended to be modular with a proximal segment that provides structural integrity to overcome gravitational challenges and a distal segment to perform the main bathing activities. The focus of this letter is on the design and control of the latter module. The design comprises of alternating tendons and pneumatics in a radial arrangement, which enables elongation, contraction, and omnidirectional bending. Additionally, a synergetic coactivation of cables and tendons in a given configuration allows for stiffness modulation, which is necessary to facilitate washing and scrubbing. The novelty of the work is twofold: 1) Three base cases of antagonistic actuation are identified that enable stiffness variation. Each category is then experimentally characterized by the application of an external force that imposes a linear displacement at the tip in both axial and lateral directions. 2) The development of a novel algorithm based on cooperative multiagent reinforcement learning that simultaneously optimizes stiffness and position. The results highlight the effectiveness of the design and control to contribute toward the development of the assistive device.
@article{ansari_multiobjective_2018,
title = {Multiobjective {Optimization} for {Stiffness} and {Position} {Control} in a {Soft} {Robot} {Arm} {Module}},
volume = {3},
doi = {10.1109/LRA.2017.2734247},
abstract = {The central concept of this letter is to develop an assistive manipulator that can automate the bathing task for elderly citizens. We propose to exploit principles of soft robotic technologies to design and control a compliant system to ensure safe human–robot interaction, a primary requirement for the task. The overall system is intended to be modular with a proximal segment that provides structural integrity to overcome gravitational challenges and a distal segment to perform the main bathing activities. The focus of this letter is on the design and control of the latter module. The design comprises of alternating tendons and pneumatics in a radial arrangement, which enables elongation, contraction, and omnidirectional bending. Additionally, a synergetic coactivation of cables and tendons in a given configuration allows for stiffness modulation, which is necessary to facilitate washing and scrubbing. The novelty of the work is twofold: 1) Three base cases of antagonistic actuation are identified that enable stiffness variation. Each category is then experimentally characterized by the application of an external force that imposes a linear displacement at the tip in both axial and lateral directions. 2) The development of a novel algorithm based on cooperative multiagent reinforcement learning that simultaneously optimizes stiffness and position. The results highlight the effectiveness of the design and control to contribute toward the development of the assistive device.},
number = {1},
journal = {IEEE Robotics and Automation Letters},
author = {Ansari, Y. and Manti, M. and Falotico, E. and Cianchetti, M. and Laschi, C.},
month = jan,
year = {2018},
keywords = {Assistive robotics, Force, Robot control, Tendons, actuators, learning (artificial intelligence), machine learning, manipulators, soft robotics},
pages = {108--115}
}Detecting Anomalous Behavior of Black-Box Services Modeled with Distance-Based Online Clustering. Gulenko, A., Schmidt, F., Acker, A., Wallschläger, M., Kao, O., & Liu, F. In 2018 IEEE 11th International Conference on Cloud Computing (CLOUD), pages 912–915, July, 2018. ISSN: 2159-6190
doi abstract bibtex
doi abstract bibtex
Reliable deployment of services is especially challenging in virtualized infrastructures, where the deep tech-nological stack and the multitude of components necessitate automatic anomaly detection and remediation mechanisms. Traditional monitoring solutions observe the system and generate alarms when the collected metrics exceed predefined thresholds. The fixed thresholds rely on expert knowledge and can lead to numerous false alarms, while abnormal behavior that spans over multiple metrics, components, or system layers, may not be detected. We propose to use an unsupervised online clustering algorithm to create a model of the normal behavior of each monitored component with minimal human interaction and no impact on the monitored system. When an anomaly is detected, a human administrator or automatic remediation system can subsequently revert the component into a normal state. An experimental evaluation resulted in a high accuracy of our approach, indicating that it is suitable for anomaly detection in productive systems.
@inproceedings{gulenko_detecting_2018,
title = {Detecting {Anomalous} {Behavior} of {Black}-{Box} {Services} {Modeled} with {Distance}-{Based} {Online} {Clustering}},
doi = {10.1109/CLOUD.2018.00134},
abstract = {Reliable deployment of services is especially challenging in virtualized infrastructures, where the deep tech-nological stack and the multitude of components necessitate automatic anomaly detection and remediation mechanisms. Traditional monitoring solutions observe the system and generate alarms when the collected metrics exceed predefined thresholds. The fixed thresholds rely on expert knowledge and can lead to numerous false alarms, while abnormal behavior that spans over multiple metrics, components, or system layers, may not be detected. We propose to use an unsupervised online clustering algorithm to create a model of the normal behavior of each monitored component with minimal human interaction and no impact on the monitored system. When an anomaly is detected, a human administrator or automatic remediation system can subsequently revert the component into a normal state. An experimental evaluation resulted in a high accuracy of our approach, indicating that it is suitable for anomaly detection in productive systems.},
booktitle = {2018 {IEEE} 11th {International} {Conference} on {Cloud} {Computing} ({CLOUD})},
author = {Gulenko, Anton and Schmidt, Florian and Acker, Alexander and Wallschläger, Marcel and Kao, Odej and Liu, Feng},
month = jul,
year = {2018},
note = {ISSN: 2159-6190},
keywords = {Anomaly detection, Cloud computing, Data collection, Data models, Measurement, Monitoring, Virtual machine monitors, anomaly detection, cloud computing, machine learning, service virtualization},
pages = {912--915},
}Machine-learning-based Dynamic IR Drop Prediction for ECO. Fang, Y., Lin, H., Su, M., Li, C., & Fang, E. J. In Proceedings of the International Conference on Computer-Aided Design, of ICCAD '18, pages 17:1–17:7, New York, NY, USA, 2018. ACM. event-place: San Diego, California
Paper doi abstract bibtex
Paper doi abstract bibtex During design signoff, many iterations of Engineer Change Order (ECO) are needed to ensure IR drop of each cell instance meets the specified limit. It is a waste of resources because repeated dynamic IR drop simulations take a very long time on very similar designs. In this work, we train a machine learning model, based on data before ECO, and predict IR drop after ECO. To increase our prediction accuracy, we propose 17 timing-aware, power-aware, and physical-aware features. Our method is scalable because the feature dimension is fixed (937), independent of design size and cell library. Also, we propose to build regional models for cell instances near IR drop violations to improves both prediction accuracy and training time. Our experiments show that our prediction correlation coefficient is 0.97 and average error is 3.0mV on a 5-million-cell industry design. Our IR drop prediction for 100K cell instances can be completed within 2 minutes. Our proposed method provides a fast IR drop prediction to speedup ECO.
@inproceedings{fang_machine-learning-based_2018,
address = {New York, NY, USA},
series = {{ICCAD} '18},
title = {Machine-learning-based {Dynamic} {IR} {Drop} {Prediction} for {ECO}},
isbn = {978-1-4503-5950-4},
url = {http://doi.acm.org/10.1145/3240765.3240823},
doi = {10.1145/3240765.3240823},
abstract = {During design signoff, many iterations of Engineer Change Order (ECO) are needed to ensure IR drop of each cell instance meets the specified limit. It is a waste of resources because repeated dynamic IR drop simulations take a very long time on very similar designs. In this work, we train a machine learning model, based on data before ECO, and predict IR drop after ECO. To increase our prediction accuracy, we propose 17 timing-aware, power-aware, and physical-aware features. Our method is scalable because the feature dimension is fixed (937), independent of design size and cell library. Also, we propose to build regional models for cell instances near IR drop violations to improves both prediction accuracy and training time. Our experiments show that our prediction correlation coefficient is 0.97 and average error is 3.0mV on a 5-million-cell industry design. Our IR drop prediction for 100K cell instances can be completed within 2 minutes. Our proposed method provides a fast IR drop prediction to speedup ECO.},
urldate = {2019-03-18},
booktitle = {Proceedings of the {International} {Conference} on {Computer}-{Aided} {Design}},
publisher = {ACM},
author = {Fang, Yen-Chun and Lin, Heng-Yi and Su, Min-Yan and Li, Chien-Mo and Fang, Eric Jia-Wei},
year = {2018},
note = {event-place: San Diego, California},
keywords = {IR drop, machine learning, power supply noise},
pages = {17:1--17:7},
}A review of machine learning in obesity. DeGregory, K., Kuiper, P., DeSilvio, T., Pleuss, J., Miller, R., Roginski, J., Fisher, C., Harness, D., Viswanath, S., Heymsfield, S., Dungan, I., & Thomas, D. Obesity Reviews, 2018.
doi abstract bibtex
doi abstract bibtex
Rich sources of obesity-related data arising from sensors, smartphone apps, electronic medical health records and insurance data can bring new insights for understanding, preventing and treating obesity. For such large datasets, machine learning provides sophisticated and elegant tools to describe, classify and predict obesity-related risks and outcomes. Here, we review machine learning methods that predict and/or classify such as linear and logistic regression, artificial neural networks, deep learning and decision tree analysis. We also review methods that describe and characterize data such as cluster analysis, principal component analysis, network science and topological data analysis. We introduce each method with a high-level overview followed by examples of successful applications. The algorithms were then applied to National Health and Nutrition Examination Survey to demonstrate methodology, utility and outcomes. The strengths and limitations of each method were also evaluated. This summary of machine learning algorithms provides a unique overview of the state of data analysis applied specifically to obesity.
@article{
title = {A review of machine learning in obesity},
type = {article},
year = {2018},
keywords = {Deep learning,National Health and Nutrition Examination Survey,machine learning,topological data analysis},
volume = {19},
id = {dc4bf489-fc6e-32a1-8ff9-11bbe7cbd6c3},
created = {2023-10-25T08:56:39.833Z},
file_attached = {false},
profile_id = {eaba325f-653b-3ee2-b960-0abd5146933e},
last_modified = {2023-10-25T08:56:39.833Z},
read = {false},
starred = {false},
authored = {true},
confirmed = {false},
hidden = {false},
private_publication = {true},
abstract = {Rich sources of obesity-related data arising from sensors, smartphone apps, electronic medical health records and insurance data can bring new insights for understanding, preventing and treating obesity. For such large datasets, machine learning provides sophisticated and elegant tools to describe, classify and predict obesity-related risks and outcomes. Here, we review machine learning methods that predict and/or classify such as linear and logistic regression, artificial neural networks, deep learning and decision tree analysis. We also review methods that describe and characterize data such as cluster analysis, principal component analysis, network science and topological data analysis. We introduce each method with a high-level overview followed by examples of successful applications. The algorithms were then applied to National Health and Nutrition Examination Survey to demonstrate methodology, utility and outcomes. The strengths and limitations of each method were also evaluated. This summary of machine learning algorithms provides a unique overview of the state of data analysis applied specifically to obesity.},
bibtype = {article},
author = {DeGregory, K.W. and Kuiper, P. and DeSilvio, T. and Pleuss, J.D. and Miller, R. and Roginski, J.W. and Fisher, C.B. and Harness, D. and Viswanath, S. and Heymsfield, S.B. and Dungan, I. and Thomas, D.M.},
doi = {10.1111/obr.12667},
journal = {Obesity Reviews},
number = {5}
}RouteNet: Routability Prediction for Mixed-size Designs Using Convolutional Neural Network. Xie, Z., Huang, Y., Fang, G., Ren, H., Fang, S., Chen, Y., & Corporation, N. In Proceedings of the International Conference on Computer-Aided Design, of ICCAD '18, pages 80:1–80:8, New York, NY, USA, 2018. ACM. event-place: San Diego, California
Paper doi abstract bibtex
Paper doi abstract bibtex Early routability prediction helps designers and tools perform preventive measures so that design rule violations can be avoided in a proactive manner. However, it is a huge challenge to have a predictor that is both accurate and fast. In this work, we study how to leverage convolutional neural network to address this challenge. The proposed method, called RouteNet, can either evaluate the overall routability of cell placement solutions without global routing or predict the locations of DRC (Design Rule Checking) hotspots. In both cases, large macros in mixed-size designs are taken into consideration. Experiments on benchmark circuits show that RouteNet can forecast overall routability with accuracy similar to that of global router while using substantially less runtime. For DRC hotspot prediction, RouteNet improves accuracy by 50% compared to global routing. It also significantly outperforms other machine learning approaches such as support vector machine and logistic regression.
@inproceedings{xie_routenet_2018,
address = {New York, NY, USA},
series = {{ICCAD} '18},
title = {{RouteNet}: {Routability} {Prediction} for {Mixed}-size {Designs} {Using} {Convolutional} {Neural} {Network}},
isbn = {978-1-4503-5950-4},
shorttitle = {{RouteNet}},
url = {http://doi.acm.org/10.1145/3240765.3240843},
doi = {10.1145/3240765.3240843},
abstract = {Early routability prediction helps designers and tools perform preventive measures so that design rule violations can be avoided in a proactive manner. However, it is a huge challenge to have a predictor that is both accurate and fast. In this work, we study how to leverage convolutional neural network to address this challenge. The proposed method, called RouteNet, can either evaluate the overall routability of cell placement solutions without global routing or predict the locations of DRC (Design Rule Checking) hotspots. In both cases, large macros in mixed-size designs are taken into consideration. Experiments on benchmark circuits show that RouteNet can forecast overall routability with accuracy similar to that of global router while using substantially less runtime. For DRC hotspot prediction, RouteNet improves accuracy by 50\% compared to global routing. It also significantly outperforms other machine learning approaches such as support vector machine and logistic regression.},
urldate = {2019-03-18},
booktitle = {Proceedings of the {International} {Conference} on {Computer}-{Aided} {Design}},
publisher = {ACM},
author = {Xie, Zhiyao and Huang, Yu-Hung and Fang, Guan-Qi and Ren, Haoxing and Fang, Shao-Yun and Chen, Yiran and Corporation, Nvidia},
year = {2018},
note = {event-place: San Diego, California},
keywords = {Computer architecture, Convolutional neural networks, Layout, Machine learning, RouteNet, Routing, Runtime, Task analysis, cell placement solutions, convolutional neural nets, convolutional neural network, design rule checking, design rule violations, electronic engineering computing, integrated circuit design, logistic regression, machine learning approaches, mixed-size designs, network routing, routability prediction, support vector machine},
pages = {80:1--80:8},
}SENATUS: An Experimental SDN/NFV Orchestrator. Troia, S., Rodriguez, A., Alvizu, R., & Maier, G. In 2018 IEEE Conference on Network Function Virtualization and Software Defined Networks, NFV-SDN 2018, 2018.
doi abstract bibtex
doi abstract bibtex
© 2018 IEEE. The fifth generation telecommunication standard (5G) will make use of novel technologies, such as Software Defined Networks (SDN) and Network Function Virtualization (NFV). New models are integrating SDN and NFV in a control plane entity responsible of the Management and Orchestration (MANO) of the whole system. This entity, acting as director of the 5G control plane, is known as Service Orchestrator. This work presents SENATUS, an experimental service orchestrator targeting research and testing environments. SENATUS implements a set of innovations to support the validation of the future network planner modules that will be integrated on the management entities of the 5G architecture. Furthermore, we present two testbed scenarios to show the capability of SENATUS to control the SDN and NFV technologies previously mentioned.
@inproceedings{
title = {SENATUS: An Experimental SDN/NFV Orchestrator},
type = {inproceedings},
year = {2018},
keywords = {Machine Learning,Network Function Virtualization,OpenFlow,Openstack,Software Defined Networking},
id = {b231d989-1f45-3374-ac2b-991b78043094},
created = {2022-12-27T17:16:58.185Z},
file_attached = {false},
profile_id = {32f51a56-0d81-30ac-a19e-d4a75485e4d6},
group_id = {96023cb7-9d23-3351-86e1-343ad61f24e6},
last_modified = {2022-12-27T17:16:58.185Z},
read = {false},
starred = {false},
authored = {false},
confirmed = {false},
hidden = {false},
private_publication = {false},
abstract = {© 2018 IEEE. The fifth generation telecommunication standard (5G) will make use of novel technologies, such as Software Defined Networks (SDN) and Network Function Virtualization (NFV). New models are integrating SDN and NFV in a control plane entity responsible of the Management and Orchestration (MANO) of the whole system. This entity, acting as director of the 5G control plane, is known as Service Orchestrator. This work presents SENATUS, an experimental service orchestrator targeting research and testing environments. SENATUS implements a set of innovations to support the validation of the future network planner modules that will be integrated on the management entities of the 5G architecture. Furthermore, we present two testbed scenarios to show the capability of SENATUS to control the SDN and NFV technologies previously mentioned.},
bibtype = {inproceedings},
author = {Troia, S. and Rodriguez, A. and Alvizu, R. and Maier, G.},
doi = {10.1109/NFV-SDN.2018.8725690},
booktitle = {2018 IEEE Conference on Network Function Virtualization and Software Defined Networks, NFV-SDN 2018}
}Online ensemble learning with abstaining classifiers for drifting and noisy data streams. Krawczyk, B. & Cano, A. Applied Soft Computing, 68:677–692, July, 2018.
Paper doi abstract bibtex
Paper doi abstract bibtex Mining data streams is among most vital contemporary topics in machine learning. Such scenario requires adaptive algorithms that are able to process constantly arriving instances, adapt to potential changes in data, use limited computational resources, as well as be robust to any atypical events that may appear. Ensemble learning has proven itself to be an effective solution, as combining learners leads to an improved predictive power, more flexible drift handling, as well as ease of being implemented in high-performance computing environments. In this paper, we propose an enhancement of popular online ensembles by augmenting them with abstaining option. Instead of relying on a traditional voting, classifiers are allowed to abstain from contributing to the final decision. Their confidence level is being monitored for each incoming instance and only learners that exceed certain threshold are selected. We introduce a dynamic and self-adapting threshold that is able to adapt to changes in the data stream, by monitoring outputs of the ensemble and allowing to exploit underlying diversity in order to efficiently anticipate drifts. Additionally, we show that forcing uncertain classifiers to abstain from making a prediction is especially useful for noisy data streams. Our proposal is a lightweight enhancement that can be applied to any online ensemble method, improving its robustness to drifts and noise. Thorough experimental analysis validated through statistical tests proves the usefulness of the proposed approach.
@article{krawczyk_online_2018,
title = {Online ensemble learning with abstaining classifiers for drifting and noisy data streams},
volume = {68},
issn = {1568-4946},
url = {http://www.sciencedirect.com/science/article/pii/S1568494617307238},
doi = {10.1016/j.asoc.2017.12.008},
abstract = {Mining data streams is among most vital contemporary topics in machine learning. Such scenario requires adaptive algorithms that are able to process constantly arriving instances, adapt to potential changes in data, use limited computational resources, as well as be robust to any atypical events that may appear. Ensemble learning has proven itself to be an effective solution, as combining learners leads to an improved predictive power, more flexible drift handling, as well as ease of being implemented in high-performance computing environments. In this paper, we propose an enhancement of popular online ensembles by augmenting them with abstaining option. Instead of relying on a traditional voting, classifiers are allowed to abstain from contributing to the final decision. Their confidence level is being monitored for each incoming instance and only learners that exceed certain threshold are selected. We introduce a dynamic and self-adapting threshold that is able to adapt to changes in the data stream, by monitoring outputs of the ensemble and allowing to exploit underlying diversity in order to efficiently anticipate drifts. Additionally, we show that forcing uncertain classifiers to abstain from making a prediction is especially useful for noisy data streams. Our proposal is a lightweight enhancement that can be applied to any online ensemble method, improving its robustness to drifts and noise. Thorough experimental analysis validated through statistical tests proves the usefulness of the proposed approach.},
language = {en},
urldate = {2020-12-12},
journal = {Applied Soft Computing},
author = {Krawczyk, Bartosz and Cano, Alberto},
month = jul,
year = {2018},
keywords = {Abstaining classifier, Concept drift, Data stream mining, Diversity, Ensemble learning, Machine learning},
pages = {677--692},
}Gaze and the Control of Foot Placement When Walking in Natural Terrain. Matthis, J. S., Yates, J. L., & Hayhoe, M. M. Current Biology, April, 2018.
Paper doi bibtex
Paper doi bibtex @article{matthis_gaze_2018,
title = {Gaze and the {Control} of {Foot} {Placement} {When} {Walking} in {Natural} {Terrain}},
volume = {0},
issn = {0960-9822},
url = {http://www.cell.com/current-biology/abstract/S0960-9822(18)30309-9},
doi = {10.1016/j.cub.2018.03.008},
language = {English},
number = {0},
urldate = {2018-04-13},
journal = {Current Biology},
author = {Matthis, Jonathan Samir and Yates, Jacob L. and Hayhoe, Mary M.},
month = apr,
year = {2018},
keywords = {eye movements, foot placement, gaze, locomotion, real-world, rough terrain, walking}
}Sleep Duration and Physical Activity Profiles Associated With Self-Reported Stroke in the United States: Application of Bayesian Belief Network Modeling Techniques. Seixas, A., A., Henclewood, D., A., Williams, S., K., Jagannathan, R., Ramos, A., Zizi, F., & Jean-Louis, G. Frontiers in Neurology, 9:534, Frontiers, 7, 2018.
Website doi abstract bibtex
Website doi abstract bibtex Introduction: Physical activity (PA) and sleep are associated with cerebrovascular disease and events like stroke. Though the interrelationships between PA, sleep, and other stroke risk factors have been studied, we are unclear about the associations of different types, frequency and duration of PA, sleep behavioral patterns (short, average and long sleep durations), within the context of stroke-related clinical, behavioral, and socio-demographic risk factors. The current study utilized Bayesian Belief Network analysis, a type of machine learning analysis, to develop profiles of physical activity (duration, intensity, and frequency) and sleep duration associated with or no history of stroke, given the influence of multiple stroke predictors and correlates. Such a model allowed us to develop a predictive classification model of stroke which can be used in post-stroke risk stratification and developing targeted stroke rehabilitation care based on an individual’s profile. Method: Analysis was based on the 2004-2013 National Health Interview Survey (n=288,888). Bayesian Belief Network analysis (BBN) was used to model the omnidirectional relationships of sleep duration and physical activity to history of stroke. Demographic, behavioral, health/medical, and psychosocial factors were considered as well as sleep duration [defined as short < 7hrs. and long ≥ 9hrs., referenced to healthy sleep (7-8 hrs.)], and intensity (moderate and vigorous) and frequency (times/week) of physical activity. Results: Of the sample, 48.1% were ≤45 years; 55.7% female; 77.4% were White; 15.9%, Black/African American; and 45.3% reported an annual income< $35K. Overall, the model had a precision index of 95.84%. We found that adults who reported 31-60 minutes of vigorous physical activity six times for the week and average sleep duration (7-8 hrs.) had the lowest stroke prevalence. Of the 36 sleep (short, average, and long sleep) and physical activity profiles we tested, 30 profiles had a self-reported stroke prevalence lower than the US national average of approximately 3.07%. Women, compared to men with the same sleep and physical activity profile, appeared to have higher self-reported stroke prevalence. We also report age differences across three groups 18-45, 46-65, and 66+ . Conclusion: Our findings indicate that several profiles of sleep duration and physical activity are associated with
@article{
title = {Sleep Duration and Physical Activity Profiles Associated With Self-Reported Stroke in the United States: Application of Bayesian Belief Network Modeling Techniques},
type = {article},
year = {2018},
keywords = {Stroke,gender,machine learning,physical activity,sleep duration},
pages = {534},
volume = {9},
websites = {https://www.frontiersin.org/article/10.3389/fneur.2018.00534/full},
month = {7},
publisher = {Frontiers},
day = {19},
id = {09dfc6a1-a77c-3ade-be08-a4dfd1608364},
created = {2018-07-22T20:13:09.567Z},
accessed = {2018-07-22},
file_attached = {false},
profile_id = {95e10851-cdf3-31de-9f82-1ab629e601b0},
group_id = {838ecfe2-7c01-38b2-970d-875a87910530},
last_modified = {2018-07-22T20:13:09.567Z},
read = {false},
starred = {false},
authored = {false},
confirmed = {false},
hidden = {false},
private_publication = {false},
abstract = {Introduction: Physical activity (PA) and sleep are associated with cerebrovascular disease and events like stroke. Though the interrelationships between PA, sleep, and other stroke risk factors have been studied, we are unclear about the associations of different types, frequency and duration of PA, sleep behavioral patterns (short, average and long sleep durations), within the context of stroke-related clinical, behavioral, and socio-demographic risk factors. The current study utilized Bayesian Belief Network analysis, a type of machine learning analysis, to develop profiles of physical activity (duration, intensity, and frequency) and sleep duration associated with or no history of stroke, given the influence of multiple stroke predictors and correlates. Such a model allowed us to develop a predictive classification model of stroke which can be used in post-stroke risk stratification and developing targeted stroke rehabilitation care based on an individual’s profile. Method: Analysis was based on the 2004-2013 National Health Interview Survey (n=288,888). Bayesian Belief Network analysis (BBN) was used to model the omnidirectional relationships of sleep duration and physical activity to history of stroke. Demographic, behavioral, health/medical, and psychosocial factors were considered as well as sleep duration [defined as short < 7hrs. and long ≥ 9hrs., referenced to healthy sleep (7-8 hrs.)], and intensity (moderate and vigorous) and frequency (times/week) of physical activity. Results: Of the sample, 48.1% were ≤45 years; 55.7% female; 77.4% were White; 15.9%, Black/African American; and 45.3% reported an annual income< $35K. Overall, the model had a precision index of 95.84%. We found that adults who reported 31-60 minutes of vigorous physical activity six times for the week and average sleep duration (7-8 hrs.) had the lowest stroke prevalence. Of the 36 sleep (short, average, and long sleep) and physical activity profiles we tested, 30 profiles had a self-reported stroke prevalence lower than the US national average of approximately 3.07%. Women, compared to men with the same sleep and physical activity profile, appeared to have higher self-reported stroke prevalence. We also report age differences across three groups 18-45, 46-65, and 66+ . Conclusion: Our findings indicate that several profiles of sleep duration and physical activity are associated with},
bibtype = {article},
author = {Seixas, Azizi A. and Henclewood, Dwayne A. and Williams, Stephen K. and Jagannathan, Ram and Ramos, Alberto and Zizi, Ferdinand and Jean-Louis, Girardin},
doi = {10.3389/fneur.2018.00534},
journal = {Frontiers in Neurology}
}2017 (5)
Deep Learning in Medical Imaging: General Overview. Lee, J., Jun, S., Cho, Y., Lee, H., Kim, G. B., Seo, J. B., & Kim, N. Korean journal of radiology, 18(4):570–584, August, 2017. Place: Korea (South)
doi abstract bibtex
doi abstract bibtex
The artificial neural network (ANN)-a machine learning technique inspired by the human neuronal synapse system-was introduced in the 1950s. However, the ANN was previously limited in its ability to solve actual problems, due to the vanishing gradient and overfitting problems with training of deep architecture, lack of computing power, and primarily the absence of sufficient data to train the computer system. Interest in this concept has lately resurfaced, due to the availability of big data, enhanced computing power with the current graphics processing units, and novel algorithms to train the deep neural network. Recent studies on this technology suggest its potentially to perform better than humans in some visual and auditory recognition tasks, which may portend its applications in medicine and healthcare, especially in medical imaging, in the foreseeable future. This review article offers perspectives on the history, development, and applications of deep learning technology, particularly regarding its applications in medical imaging.
@article{lee_deep_2017,
title = {Deep {Learning} in {Medical} {Imaging}: {General} {Overview}.},
volume = {18},
issn = {2005-8330 1229-6929},
doi = {10.3348/kjr.2017.18.4.570},
abstract = {The artificial neural network (ANN)-a machine learning technique inspired by the human neuronal synapse system-was introduced in the 1950s. However, the ANN was previously limited in its ability to solve actual problems, due to the vanishing gradient and overfitting problems with training of deep architecture, lack of computing power, and primarily the absence of sufficient data to train the computer system. Interest in this concept has lately resurfaced, due to the availability of big data, enhanced computing power with the current graphics processing units, and novel algorithms to train the deep neural network. Recent studies on this technology suggest its potentially to perform better than humans in some visual and auditory recognition tasks, which may portend its applications in medicine and healthcare, especially in medical imaging, in the foreseeable future. This review article offers perspectives on the history, development, and applications of deep learning technology, particularly regarding its applications in medical imaging.},
language = {eng},
number = {4},
journal = {Korean journal of radiology},
author = {Lee, June-Goo and Jun, Sanghoon and Cho, Young-Won and Lee, Hyunna and Kim, Guk Bae and Seo, Joon Beom and Kim, Namkug},
month = aug,
year = {2017},
pmid = {28670152},
pmcid = {PMC5447633},
note = {Place: Korea (South)},
keywords = {*Algorithms, *Neural Networks, Computer, Artificial intelligence, Computer-aided, Convolutional neural network, Humans, Image Processing, Computer-Assisted, Knee/diagnostic imaging, Machine learning, Magnetic Resonance Imaging, Optical Imaging/methods, Precision medicine, Radiology, Recurrent Neural Network},
pages = {570--584},
}BrainNetCNN: Artificial Convolutional Neural Networks for Connectomes. Kawahara, J., Brown, C. J., Miller, S., Booth, B. G., Chau, V., Grunau, R., Zwicker, J., & Hamarneh, G. In 2nd Annual Health Technology Symposium, Vancouver, Canada, pages 1, 2017.
bibtex
bibtex
@INPROCEEDINGS{healthtech2017c,
OPTADDRESS = {},
AUTHOR = {Jeremy Kawahara and Colin J. Brown and Steven Miller and
Brian G. Booth and Vann Chau and Ruth Grunau and Jill Zwicker and
Ghassan Hamarneh},
BOOKTITLE = {2nd Annual Health Technology Symposium, Vancouver, Canada},
OPTEDITOR = {},
OPTMONTH = {},
OPTNOTE = {},
OPTNUMBER = {},
OPTORGANIZATION = {},
PAGES = {1},
OPTPUBLISHER = {},
OPTSERIES = {},
TITLE = {BrainNetCNN: Artificial Convolutional Neural Networks
for Connectomes},
OPTVOLUME = {},
YEAR = {2017},
OPTABSTRACT = {},
OPTDOI = {},
OPTISBN = {},
OPTISSN = {},
KEYWORDS = {Diffusion MRI, Neurodevelopment, Connectome,
Machine Learning, Deep Learning, Classification},
OPTURL = {},
OPTURL-PUBLISHER = {},
PDF = {http://www.cs.sfu.ca/~hamarneh/ecopy/healthtech2017c.pdf}
}Deep learning for computational chemistry. Goh, G. B, Hodas, N. O, & Vishnu, A. Journal of Computational Chemistry, 38(16):1291–1307, 2017.
Paper doi bibtex
Paper doi bibtex @article{goh_deep_2017,
title = {Deep learning for computational chemistry},
volume = {38},
issn = {1096-987X},
url = {http://dx.doi.org/10.1002/jcc.24764},
doi = {10.1002/jcc.24764},
number = {16},
journal = {Journal of Computational Chemistry},
author = {Goh, Garrett B and Hodas, Nathan O and Vishnu, Abhinav},
year = {2017},
keywords = {artificial intelligence, cheminformatics, deep learning, machine learning, materials genome, molecular modeling, protein structure prediction, quantitative structure activity relationship, quantum chemistry, toxicology},
pages = {1291--1307},
}The Matter Simulation (R)evolution. Aspuru-Guzik, A., Lindh, R., & Reiher, M. November, 2017.
Paper doi abstract bibtex
Paper doi abstract bibtex To date, the program for the development of methods and models for atomistic and continuum simulation directed toward chemicals and materials has reached an incredible degree of sophistication and maturity. Currently, one can witness an increasingly rapid emergence of advances in computing, artificial intelligence, and robotics. This drives us to consider the future of computer simulation of matter from the molecular to the human length and time scales in a radical way that deliberately dares to go beyond the foreseeable next steps in any given discipline. This perspective article presents a view on this future development that we believe is likely to become a reality during our lifetime.
@misc{aspuru-guzik_matter_2017,
title = {The {Matter} {Simulation} ({R})evolution},
url = {https://chemrxiv.org/articles/The_Matter_Simulation_R_evolution/5616115},
abstract = {To date, the program for the development of methods and models for atomistic and continuum simulation directed toward chemicals and materials has reached an incredible degree of sophistication and maturity. Currently, one can witness an increasingly rapid emergence of advances in computing, artificial intelligence, and robotics. This drives us to consider the future of computer simulation of matter from the molecular to the human length and time scales in a radical way that deliberately dares to go beyond the foreseeable next steps in any given discipline. This perspective article presents a view on this future development that we believe is likely to become a reality during our lifetime.},
author = {Aspuru-Guzik, Alan and Lindh, Roland and Reiher, Markus},
month = nov,
year = {2017},
doi = {10.26434/chemrxiv.5616115.v1},
keywords = {Artificial Intelligence, Augmented Reality, Computational Chemistry, Grand Challenges, Machine Learning, Materials Science, Perspective, Quantum Computing, Quantum Information, Quantum Simulation, Robotics, Theoretical chemistry, Turing Test},
}Training an Actor-Critic Reinforcement Learning Controller for Arm Movement Using Human-Generated Rewards. Jagodnik, K. M., Thomas, P. S., van den Bogert, A. J., Branicky, M. S., & Kirsch, R. F. IEEE transactions on neural systems and rehabilitation engineering: a publication of the IEEE Engineering in Medicine and Biology Society, 25(10):1892–1905, 2017.
doi abstract bibtex
doi abstract bibtex
Functional Electrical Stimulation (FES) employs neuroprostheses to apply electrical current to the nerves and muscles of individuals paralyzed by spinal cord injury to restore voluntary movement. Neuroprosthesis controllers calculate stimulation patterns to produce desired actions. To date, no existing controller is able to efficiently adapt its control strategy to the wide range of possible physiological arm characteristics, reaching movements, and user preferences that vary over time. Reinforcement learning (RL) is a control strategy that can incorporate human reward signals as inputs to allow human users to shape controller behavior. In this paper, ten neurologically intact human participants assigned subjective numerical rewards to train RL controllers, evaluating animations of goal-oriented reaching tasks performed using a planar musculoskeletal human arm simulation. The RL controller learning achieved using human trainers was compared with learning accomplished using human-like rewards generated by an algorithm; metrics included success at reaching the specified target; time required to reach the target; and target overshoot. Both sets of controllers learned efficiently and with minimal differences, significantly outperforming standard controllers. Reward positivity and consistency were found to be unrelated to learning success. These results suggest that human rewards can be used effectively to train RL-based FES controllers.
@article{jagodnik_training_2017,
title = {Training an {Actor}-{Critic} {Reinforcement} {Learning} {Controller} for {Arm} {Movement} {Using} {Human}-{Generated} {Rewards}},
volume = {25},
copyright = {All rights reserved},
issn = {1558-0210},
doi = {10.1109/TNSRE.2017.2700395},
abstract = {Functional Electrical Stimulation (FES) employs neuroprostheses to apply electrical current to the nerves and muscles of individuals paralyzed by spinal cord injury to restore voluntary movement. Neuroprosthesis controllers calculate stimulation patterns to produce desired actions. To date, no existing controller is able to efficiently adapt its control strategy to the wide range of possible physiological arm characteristics, reaching movements, and user preferences that vary over time. Reinforcement learning (RL) is a control strategy that can incorporate human reward signals as inputs to allow human users to shape controller behavior. In this paper, ten neurologically intact human participants assigned subjective numerical rewards to train RL controllers, evaluating animations of goal-oriented reaching tasks performed using a planar musculoskeletal human arm simulation. The RL controller learning achieved using human trainers was compared with learning accomplished using human-like rewards generated by an algorithm; metrics included success at reaching the specified target; time required to reach the target; and target overshoot. Both sets of controllers learned efficiently and with minimal differences, significantly outperforming standard controllers. Reward positivity and consistency were found to be unrelated to learning success. These results suggest that human rewards can be used effectively to train RL-based FES controllers.},
language = {eng},
number = {10},
journal = {IEEE transactions on neural systems and rehabilitation engineering: a publication of the IEEE Engineering in Medicine and Biology Society},
author = {Jagodnik, Kathleen M. and Thomas, Philip S. and van den Bogert, Antonie J. and Branicky, Michael S. and Kirsch, Robert F.},
year = {2017},
pmid = {28475063},
pmcid = {PMC7523734},
keywords = {Adult, Algorithms, Arm, Artificial Intelligence, Biomechanical Phenomena, Electric Stimulation Therapy, Female, Healthy Volunteers, Humans, Learning, Machine Learning, Male, Motor Skills, Movement, Neural Networks, Computer, Neural Prostheses, Reinforcement, Psychology, Reward, Shoulder},
pages = {1892--1905},
}2016 (7)
S/HIC: Robust Identification of Soft and Hard Sweeps Using Machine Learning. Schrider, D. R. & Kern, A. D. PLOS Genetics, 12(3):e1005928, March, 2016.
Paper doi abstract bibtex
Paper doi abstract bibtex Detecting the targets of adaptive natural selection from whole genome sequencing data is a central problem for population genetics. However, to date most methods have shown sub-optimal performance under realistic demographic scenarios. Moreover, over the past decade there has been a renewed interest in determining the importance of selection from standing variation in adaptation of natural populations, yet very few methods for inferring this model of adaptation at the genome scale have been introduced. Here we introduce a new method, S/HIC, which uses supervised machine learning to precisely infer the location of both hard and soft selective sweeps. We show that S/HIC has unrivaled accuracy for detecting sweeps under demographic histories that are relevant to human populations, and distinguishing sweeps from linked as well as neutrally evolving regions. Moreover, we show that S/HIC is uniquely robust among its competitors to model misspecification. Thus, even if the true demographic model of a population differs catastrophically from that specified by the user, S/HIC still retains impressive discriminatory power. Finally, we apply S/HIC to the case of resequencing data from human chromosome 18 in a European population sample, and demonstrate that we can reliably recover selective sweeps that have been identified earlier using less specific and sensitive methods.
@article{schrider_s/hic:_2016,
title = {S/{HIC}: {Robust} {Identification} of {Soft} and {Hard} {Sweeps} {Using} {Machine} {Learning}},
volume = {12},
issn = {1553-7404},
shorttitle = {S/{HIC}},
url = {https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1005928},
doi = {10.1371/journal.pgen.1005928},
abstract = {Detecting the targets of adaptive natural selection from whole genome sequencing data is a central problem for population genetics. However, to date most methods have shown sub-optimal performance under realistic demographic scenarios. Moreover, over the past decade there has been a renewed interest in determining the importance of selection from standing variation in adaptation of natural populations, yet very few methods for inferring this model of adaptation at the genome scale have been introduced. Here we introduce a new method, S/HIC, which uses supervised machine learning to precisely infer the location of both hard and soft selective sweeps. We show that S/HIC has unrivaled accuracy for detecting sweeps under demographic histories that are relevant to human populations, and distinguishing sweeps from linked as well as neutrally evolving regions. Moreover, we show that S/HIC is uniquely robust among its competitors to model misspecification. Thus, even if the true demographic model of a population differs catastrophically from that specified by the user, S/HIC still retains impressive discriminatory power. Finally, we apply S/HIC to the case of resequencing data from human chromosome 18 in a European population sample, and demonstrate that we can reliably recover selective sweeps that have been identified earlier using less specific and sensitive methods.},
language = {en},
number = {3},
urldate = {2019-03-01TZ},
journal = {PLOS Genetics},
author = {Schrider, Daniel R. and Kern, Andrew D.},
month = mar,
year = {2016},
keywords = {Decision trees, Europe, Genomics statistics, Haplotypes, Machine learning, Population genetics, Population size, Simulation and modeling},
pages = {e1005928}
}Machine Learning is Fun! Part 3: Deep Learning and Convolutional Neural Networks. Geitgey, A. June, 2016.
Paper abstract bibtex
Paper abstract bibtex Update: This article is part of a series. Check out the full series: Part 1, Part 2, Part 3, Part 4, Part 5, Part 6, Part 7 and Part 8!
@misc{geitgey_machine_2016-1,
title = {Machine {Learning} is {Fun}! {Part} 3: {Deep} {Learning} and {Convolutional} {Neural} {Networks}},
shorttitle = {Machine {Learning} is {Fun}! {Part} 3},
url = {https://medium.com/@ageitgey/machine-learning-is-fun-part-3-deep-learning-and-convolutional-neural-networks-f40359318721},
abstract = {Update: This article is part of a series. Check out the full series: Part 1, Part 2, Part 3, Part 4, Part 5, Part 6, Part 7 and Part 8!},
urldate = {2018-05-03TZ},
journal = {Medium},
author = {Geitgey, Adam},
month = jun,
year = {2016},
keywords = {\#Introductory Overview, \#KeyReference, AI=Artificial Intelligence, Algorithms, CNN=Convolutional Neural Network, Classification, Computer Science, Computer Science - Computer Vision and Pattern Recognition, Computer Vision, DNN=Deep Neural Networks, Data Science, Deep Learning, GPU Computing, Image Analysis, Informatics, Machine Learning, Machine Learning Applications, Machine Learning Frameworks, Max Pooling Algorithm, Neural Networks, Python (Computer Programming), Sliding Input Window Technique, Source Code, Statistics - Machine Learning, TensorFlow, Working Code}
}Discovering Biosignatures of Cav1 Domains: Computational Methods for Super-resolution Microscopy. Khater, I. M., Meng, F., Nabi, I. R., & Hamarneh, G. In LSI Imaging Super-resolution Mini-symposium, Vancouver, Canada, pages 1, 2016.
bibtex
bibtex
@INPROCEEDINGS{superres2016,
OPTADDRESS = {},
AUTHOR = {Ismail M. Khater and Fanrui Meng and Ivan Robert Nabi and
Ghassan Hamarneh},
BOOKTITLE = {LSI Imaging Super-resolution Mini-symposium, Vancouver, Canada},
OPTEDITOR = {},
OPTMONTH = {},
OPTNOTE = {},
OPTNUMBER = {},
OPTORGANIZATION = {},
PAGES = {1},
OPTPUBLISHER = {},
OPTSERIES = {},
TITLE = {Discovering Biosignatures of Cav1 Domains: Computational
Methods for Super-resolution Microscopy},
OPTVOLUME = {},
YEAR = {2016},
OPTABSTRACT = {},
OPTDOI = {},
OPTISBN = {},
OPTISSN = {},
KEYWORDS = {Super Resolution Microscopy,
Single Molecule Localization Microscopy,
Network Modelling and Analysis, Machine Learning},
OPTURL = {},
OPTURL-PUBLISHER = {},
PDF = {http://www.cs.sfu.ca/~hamarneh/ecopy/superres2016.pdf}
}Classifying smoking urges via machine learning. Dumortier, A., Beckjord, E., Shiffman, S., & Sejdić, E. Computer Methods and Programs in Biomedicine, 137:203-213, Elsevier Ireland Ltd, 12, 2016.
Website abstract bibtex
Website abstract bibtex Background and objective Smoking is the largest preventable cause of death and diseases in the developed world, and advances in modern electronics and machine learning can help us deliver real-time intervention to smokers in novel ways. In this paper, we examine different machine learning approaches to use situational features associated with having or not having urges to smoke during a quit attempt in order to accurately classify high-urge states. Methods To test our machine learning approaches, specifically, Bayes, discriminant analysis and decision tree learning methods, we used a dataset collected from over 300 participants who had initiated a quit attempt. The three classification approaches are evaluated observing sensitivity, specificity, accuracy and precision. Results The outcome of the analysis showed that algorithms based on feature selection make it possible to obtain high classification rates with only a few features selected from the entire dataset. The classification tree method outperformed the naive Bayes and discriminant analysis methods, with an accuracy of the classifications up to 86%. These numbers suggest that machine learning may be a suitable approach to deal with smoking cessation matters, and to predict smoking urges, outlining a potential use for mobile health applications. Conclusions In conclusion, machine learning classifiers can help identify smoking situations, and the search for the best features and classifier parameters significantly improves the algorithms’ performance. In addition, this study also supports the usefulness of new technologies in improving the effect of smoking cessation interventions, the management of time and patients by therapists, and thus the optimization of available health care resources. Future studies should focus on providing more adaptive and personalized support to people who really need it, in a minimum amount of time by developing novel expert systems capable of delivering real-time interventions.
@article{
title = {Classifying smoking urges via machine learning},
type = {article},
year = {2016},
identifiers = {[object Object]},
keywords = {Feature selection,Machine learning,Smoking cessation,Smoking urges,Supervised learning},
pages = {203-213},
volume = {137},
websites = {/pmc/articles/PMC5289882/?report=abstract,https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5289882/},
month = {12},
publisher = {Elsevier Ireland Ltd},
day = {1},
id = {63de798d-05f0-3b29-8c92-9758a2c070c2},
created = {2020-09-25T08:37:57.426Z},
accessed = {2020-09-22},
file_attached = {false},
profile_id = {9b09ea17-50dc-3505-8d03-5f32efb22754},
group_id = {5bb643b5-536a-34eb-a989-ad28d02bdb1a},
last_modified = {2020-09-25T08:37:57.426Z},
read = {false},
starred = {false},
authored = {false},
confirmed = {false},
hidden = {false},
private_publication = {false},
abstract = {Background and objective Smoking is the largest preventable cause of death and diseases in the developed world, and advances in modern electronics and machine learning can help us deliver real-time intervention to smokers in novel ways. In this paper, we examine different machine learning approaches to use situational features associated with having or not having urges to smoke during a quit attempt in order to accurately classify high-urge states. Methods To test our machine learning approaches, specifically, Bayes, discriminant analysis and decision tree learning methods, we used a dataset collected from over 300 participants who had initiated a quit attempt. The three classification approaches are evaluated observing sensitivity, specificity, accuracy and precision. Results The outcome of the analysis showed that algorithms based on feature selection make it possible to obtain high classification rates with only a few features selected from the entire dataset. The classification tree method outperformed the naive Bayes and discriminant analysis methods, with an accuracy of the classifications up to 86%. These numbers suggest that machine learning may be a suitable approach to deal with smoking cessation matters, and to predict smoking urges, outlining a potential use for mobile health applications. Conclusions In conclusion, machine learning classifiers can help identify smoking situations, and the search for the best features and classifier parameters significantly improves the algorithms’ performance. In addition, this study also supports the usefulness of new technologies in improving the effect of smoking cessation interventions, the management of time and patients by therapists, and thus the optimization of available health care resources. Future studies should focus on providing more adaptive and personalized support to people who really need it, in a minimum amount of time by developing novel expert systems capable of delivering real-time interventions.},
bibtype = {article},
author = {Dumortier, Antoine and Beckjord, Ellen and Shiffman, Saul and Sejdić, Ervin},
journal = {Computer Methods and Programs in Biomedicine}
}Applying GIS and Machine Learning Methods to Twitter Data for Multiscale Surveillance of Influenza. Allen, C., Tsou, M., Aslam, A., Nagel, A., & Gawron, J. PLOS ONE, 11(7):e0157734, July, 2016. 00000 Publisher: Public Library of Science
Paper doi abstract bibtex
Paper doi abstract bibtex Traditional methods for monitoring influenza are haphazard and lack fine-grained details regarding the spatial and temporal dynamics of outbreaks. Twitter gives researchers and public health officials an opportunity to examine the spread of influenza in real-time and at multiple geographical scales. In this paper, we introduce an improved framework for monitoring influenza outbreaks using the social media platform Twitter. Relying upon techniques from geographic information science (GIS) and data mining, Twitter messages were collected, filtered, and analyzed for the thirty most populated cities in the United States during the 2013–2014 flu season. The results of this procedure are compared with national, regional, and local flu outbreak reports, revealing a statistically significant correlation between the two data sources. The main contribution of this paper is to introduce a comprehensive data mining process that enhances previous attempts to accurately identify tweets related to influenza. Additionally, geographical information systems allow us to target, filter, and normalize Twitter messages.
@article{allen_applying_2016,
title = {Applying {GIS} and {Machine} {Learning} {Methods} to {Twitter} {Data} for {Multiscale} {Surveillance} of {Influenza}},
volume = {11},
issn = {1932-6203},
url = {https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0157734},
doi = {10.1371/journal.pone.0157734},
abstract = {Traditional methods for monitoring influenza are haphazard and lack fine-grained details regarding the spatial and temporal dynamics of outbreaks. Twitter gives researchers and public health officials an opportunity to examine the spread of influenza in real-time and at multiple geographical scales. In this paper, we introduce an improved framework for monitoring influenza outbreaks using the social media platform Twitter. Relying upon techniques from geographic information science (GIS) and data mining, Twitter messages were collected, filtered, and analyzed for the thirty most populated cities in the United States during the 2013–2014 flu season. The results of this procedure are compared with national, regional, and local flu outbreak reports, revealing a statistically significant correlation between the two data sources. The main contribution of this paper is to introduce a comprehensive data mining process that enhances previous attempts to accurately identify tweets related to influenza. Additionally, geographical information systems allow us to target, filter, and normalize Twitter messages.},
language = {en},
number = {7},
urldate = {2020-02-27},
journal = {PLOS ONE},
author = {Allen, Chris and Tsou, Ming-Hsiang and Aslam, Anoshe and Nagel, Anna and Gawron, Jean-Mark},
month = jul,
year = {2016},
note = {00000
Publisher: Public Library of Science},
keywords = {Geographic information systems, Influenza, Machine learning, Machine learning algorithms, Public and occupational health, Social media, Support vector machines, Twitter},
pages = {e0157734},
}Problematic internet use (PIU): Associations with the impulsive-compulsive spectrum. An application of machine learning in psychiatry. Ioannidis, K., Chamberlain, S., R., Treder, M., S., Kiraly, F., Leppink, E., W., Redden, S., A., Stein, D., J., Lochner, C., & Grant, J., E. Journal of Psychiatric Research, 83:94-102, 2016.
Website abstract bibtex
Website abstract bibtex Problematic internet use is common, functionally impairing, and in need of further study. Its relationship with obsessive-compulsive and impulsive disorders is unclear. Our objective was to evaluate whether problematic internet use can be predicted from recognised forms of impulsive and compulsive traits and symptomatology. We recruited volunteers aged 18 and older using media advertisements at two sites (Chicago USA, and Stellenbosch, South Africa) to complete an extensive online survey. State-of-the-art out-of-sample evaluation of machine learning predictive models was used, which included Logistic Regression, Random Forests and Naïve Bayes. Problematic internet use was identified using the Internet Addiction Test (IAT). 2006 complete cases were analysed, of whom 181 (9.0%) had moderate/severe problematic internet use. Using Logistic Regression and Naïve Bayes we produced a classification prediction with a receiver operating characteristic area under the curve (ROC-AUC) of 0.83 (SD 0.03) whereas using a Random Forests algorithm the prediction ROC-AUC was 0.84 (SD 0.03) [all three models superior to baseline models p < 0.0001]. The models showed robust transfer between the study sites in all validation sets [p < 0.0001]. Prediction of problematic internet use was possible using specific measures of impulsivity and compulsivity in a population of volunteers. Moreover, this study offers proof-of-concept in support of using machine learning in psychiatry to demonstrate replicability of results across geographically and culturally distinct settings.
@article{
title = {Problematic internet use (PIU): Associations with the impulsive-compulsive spectrum. An application of machine learning in psychiatry},
type = {article},
year = {2016},
identifiers = {[object Object]},
keywords = {ADHD,Adolescent,Adult,Aged,Aged, 80 and over,Behavior, Addictive,Compulsive Behavior,Compulsivity,Female,Humans,Impulsivity,Internet,Internet use,Machine Learning,Machine learning,Male,Middle Aged,OCD,Obsessive-Compulsive Disorder,Online Systems,Predictive Value of Tests,Psychiatry,ROC Curve,Reproducibility of Results,South Africa,Surveys and Questionnaires,United States,Young Adult},
pages = {94-102},
volume = {83},
websites = {http://files/1161/Ioannidis et al. - 2016 - Problematic internet use (PIU) Associations with .pdf,http://www.ncbi.nlm.nih.gov/pubmed/27580487},
id = {018be67c-9f6d-3317-bdff-f0e36f34becb},
created = {2020-09-17T09:27:55.706Z},
file_attached = {false},
profile_id = {20f87055-ac78-3c65-9cf5-216a3558d16a},
group_id = {14ca8526-77d5-34fd-89de-e48cae5e6ee2},
last_modified = {2020-09-17T09:27:55.706Z},
read = {false},
starred = {false},
authored = {false},
confirmed = {true},
hidden = {false},
source_type = {JOUR},
language = {eng},
private_publication = {false},
abstract = {Problematic internet use is common, functionally impairing, and in need of further study. Its relationship with obsessive-compulsive and impulsive disorders is unclear. Our objective was to evaluate whether problematic internet use can be predicted from recognised forms of impulsive and compulsive traits and symptomatology. We recruited volunteers aged 18 and older using media advertisements at two sites (Chicago USA, and Stellenbosch, South Africa) to complete an extensive online survey. State-of-the-art out-of-sample evaluation of machine learning predictive models was used, which included Logistic Regression, Random Forests and Naïve Bayes. Problematic internet use was identified using the Internet Addiction Test (IAT). 2006 complete cases were analysed, of whom 181 (9.0%) had moderate/severe problematic internet use. Using Logistic Regression and Naïve Bayes we produced a classification prediction with a receiver operating characteristic area under the curve (ROC-AUC) of 0.83 (SD 0.03) whereas using a Random Forests algorithm the prediction ROC-AUC was 0.84 (SD 0.03) [all three models superior to baseline models p < 0.0001]. The models showed robust transfer between the study sites in all validation sets [p < 0.0001]. Prediction of problematic internet use was possible using specific measures of impulsivity and compulsivity in a population of volunteers. Moreover, this study offers proof-of-concept in support of using machine learning in psychiatry to demonstrate replicability of results across geographically and culturally distinct settings.},
bibtype = {article},
author = {Ioannidis, Konstantinos and Chamberlain, Samuel R and Treder, Matthias S and Kiraly, Franz and Leppink, Eric W and Redden, Sarah A and Stein, Dan J and Lochner, Christine and Grant, Jon E},
journal = {Journal of Psychiatric Research}
}Clinically-Inspired Automatic Classification of Ovarian Carcinoma Subtypes. BenTaieb, A., Nosrati, M., Li-Chang, H., Huntsman, D., & Hamarneh, G. Journal of Pathology Informatics, 7(1):1-28, 2016.
doi bibtex
doi bibtex
@ARTICLE{jpi2016,
AUTHOR = {Aicha BenTaieb and Masoud Nosrati and Hector Li-Chang and
David Huntsman and Ghassan Hamarneh},
JOURNAL = {Journal of Pathology Informatics},
OPTMONTH = {},
OPTNOTE = {},
NUMBER = {1},
PAGES = {1-28},
TITLE = {Clinically-Inspired Automatic Classification of Ovarian
Carcinoma Subtypes},
VOLUME = {7},
YEAR = {2016},
OPTABSTRACT = {},
DOI = {10.4103/2153-3539.186899},
OPTISBN = {},
OPTISSN = {},
KEYWORDS = {Color, Machine Learning, Microscopy},
OPTURL = {},
OPTURL-PUBLISHER = {},
PDF = {http://www.cs.sfu.ca/~hamarneh/ecopy/jpi2016.pdf}
}2015 (5)
Predicting students' happiness from physiology, phone, mobility, and behavioral data. Jaques, N., Taylor, S., Azaria, A., Ghandeharioun, A., Sano, A., & Picard, R. In 2015 International Conference on Affective Computing and Intelligent Interaction, ACII 2015, volume 2015, pages 222-228, 12, 2015. Institute of Electrical and Electronics Engineers Inc..
Website abstract bibtex
Website abstract bibtex In order to model students' happiness, we apply machine learning methods to data collected from undergrad students monitored over the course of one month each. The data collected include physiological signals, location, smartphone logs, and survey responses to behavioral questions. Each day, participants reported their wellbeing on measures including stress, health, and happiness. Because of the relationship between happiness and depression, modeling happiness may help us to detect individuals who are at risk of depression and guide interventions to help them. We are also interested in how behavioral factors (such as sleep and social activity) affect happiness positively and negatively. A variety of machine learning and feature selection techniques are compared, including Gaussian Mixture Models and ensemble classification. We achieve 70% classification accuracy of self-reported happiness on held-out test data.
@inproceedings{
title = {Predicting students' happiness from physiology, phone, mobility, and behavioral data},
type = {inproceedings},
year = {2015},
identifiers = {[object Object]},
keywords = {happiness,machine learning,wellbeing},
pages = {222-228},
volume = {2015},
websites = {/pmc/articles/PMC5431070/?report=abstract,https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5431070/},
month = {12},
publisher = {Institute of Electrical and Electronics Engineers Inc.},
day = {2},
id = {6dc481bd-bf62-38e1-a599-c7b28f89ddfb},
created = {2020-09-25T08:37:57.448Z},
accessed = {2020-09-22},
file_attached = {false},
profile_id = {9b09ea17-50dc-3505-8d03-5f32efb22754},
group_id = {5bb643b5-536a-34eb-a989-ad28d02bdb1a},
last_modified = {2020-09-25T08:37:57.448Z},
read = {false},
starred = {false},
authored = {false},
confirmed = {false},
hidden = {false},
private_publication = {false},
abstract = {In order to model students' happiness, we apply machine learning methods to data collected from undergrad students monitored over the course of one month each. The data collected include physiological signals, location, smartphone logs, and survey responses to behavioral questions. Each day, participants reported their wellbeing on measures including stress, health, and happiness. Because of the relationship between happiness and depression, modeling happiness may help us to detect individuals who are at risk of depression and guide interventions to help them. We are also interested in how behavioral factors (such as sleep and social activity) affect happiness positively and negatively. A variety of machine learning and feature selection techniques are compared, including Gaussian Mixture Models and ensemble classification. We achieve 70% classification accuracy of self-reported happiness on held-out test data.},
bibtype = {inproceedings},
author = {Jaques, Natasha and Taylor, Sara and Azaria, Asaph and Ghandeharioun, Asma and Sano, Akane and Picard, Rosalind},
booktitle = {2015 International Conference on Affective Computing and Intelligent Interaction, ACII 2015}
}Probabilistic event calculus for event recognition. Skarlatidis, A., Paliouras, G., Artikis, A., & Vouros, G. A. ACM Trans. Comput. Logic, 16(2):11:1--11:37, ACM, New York, NY, USA, feb, 2015.
Paper doi bibtex
Paper doi bibtex @Article{Skarlatidis_2015_17615,
author = {Skarlatidis, Anastasios and Paliouras, Georgios and Artikis, Alexander and Vouros, George A.},
address = {New York, NY, USA},
journal = {ACM Trans. Comput. Logic},
month = {feb},
number = {2},
pages = {11:1--11:37},
publisher = {ACM},
title = {Probabilistic event calculus for event recognition},
volume = {16},
year = {2015},
issn = {1529-3785},
keywords = {Events, machine learning, probabilistic inference, uncertainty},
url = {http://doi.acm.org/10.1145/2699916},
doi = {10.1145/2699916},
title_with_no_special_chars = {Probabilistic Event Calculus for Event Recognition}
}Bayesian kernel machine regression for estimating the health effects of multi-pollutant mixtures. Bobb, J. F., Valeri, L., Claus Henn, B., Christiani, D. C., Wright, R. O., Mazumdar, M., Godleski, J. J., & Coull, B. A. Biostatistics (Oxford, England), 16(3):493–508, September, 2015. Publisher: Biostatistics
Paper doi abstract bibtex
Paper doi abstract bibtex Because humans are invariably exposed to complex chemical mixtures, estimating the health effects of multi-pollutant exposures is of critical concern in environmental epidemiology, and to regulatory agencies such as the U.S. Environmental Protection Agency. However, most health effects studies focus on single agents or consider simple two-way interaction models, in part because we lack the statistical methodology to more realistically capture the complexity of mixed exposures. We introduce Bayesian kernel machine regression (BKMR) as a new approach to study mixtures, in which the health outcome is regressed on a flexible function of the mixture (e.g. air pollution or toxic waste) components that is specified using a kernel function. In high-dimensional settings, a novel hierarchical variable selection approach is incorporated to identify important mixture components and account for the correlated structure of the mixture. Simulation studies demonstrate the success of BKMR in estimating the exposure-response function and in identifying the individual components of the mixture responsible for health effects. We demonstrate the features of the method through epidemiology and toxicology applications.
@article{bobb_bayesian_2015,
title = {Bayesian kernel machine regression for estimating the health effects of multi-pollutant mixtures},
volume = {16},
issn = {1468-4357},
url = {https://pubmed.ncbi.nlm.nih.gov/25532525/},
doi = {10.1093/BIOSTATISTICS/KXU058},
abstract = {Because humans are invariably exposed to complex chemical mixtures, estimating the health effects of multi-pollutant exposures is of critical concern in environmental epidemiology, and to regulatory agencies such as the U.S. Environmental Protection Agency. However, most health effects studies focus on single agents or consider simple two-way interaction models, in part because we lack the statistical methodology to more realistically capture the complexity of mixed exposures. We introduce Bayesian kernel machine regression (BKMR) as a new approach to study mixtures, in which the health outcome is regressed on a flexible function of the mixture (e.g. air pollution or toxic waste) components that is specified using a kernel function. In high-dimensional settings, a novel hierarchical variable selection approach is incorporated to identify important mixture components and account for the correlated structure of the mixture. Simulation studies demonstrate the success of BKMR in estimating the exposure-response function and in identifying the individual components of the mixture responsible for health effects. We demonstrate the features of the method through epidemiology and toxicology applications.},
number = {3},
urldate = {2021-12-16},
journal = {Biostatistics (Oxford, England)},
author = {Bobb, Jennifer F. and Valeri, Linda and Claus Henn, Birgit and Christiani, David C. and Wright, Robert O. and Mazumdar, Maitreyi and Godleski, John J. and Coull, Brent A.},
month = sep,
year = {2015},
pmid = {25532525},
note = {Publisher: Biostatistics},
keywords = {Animals, Bangladesh, Bayes Theorem*, Biostatistics, Brent A Coull, Child, Developmental Disabilities / etiology, Dogs, Environmental Health / statistics \& numerical data, Environmental Pollutants / adverse effects*, Extramural, Female, Hemodynamics / drug effects, Humans, Infant, Jennifer F Bobb, Linda Valeri, MEDLINE, Machine Learning, Metals / adverse effects, Models, N.I.H., NCBI, NIH, NLM, National Center for Biotechnology Information, National Institutes of Health, National Library of Medicine, Neurodevelopmental Disorders / etiology, Non-P.H.S., Normal Distribution, PMC5963470, Pregnancy, Preschool, PubMed Abstract, Regression Analysis, Research Support, Statistical, U.S. Gov't, doi:10.1093/biostatistics/kxu058, pmid:25532525},
pages = {493--508},
}Grand Challenge Veterinary Imaging: Technology, Science, and Communication. McEvoy, F., J. Frontiers in veterinary science, 2:38, 9, 2015.
bibtex
bibtex
@article{
title = {Grand Challenge Veterinary Imaging: Technology, Science, and Communication},
type = {article},
year = {2015},
identifiers = {[object Object]},
keywords = {clinical veterinary radiology,machine learning,teleradiology,translational models,veterinary imaging},
pages = {38},
volume = {2},
month = {9},
day = {30},
city = {Department of Veterinary Clinical and Animal Sciences, University of Copenhagen , Copenhagen , Denmark.},
id = {cb924eb7-534d-3aee-823a-b93631ee0c31},
created = {2016-09-06T13:34:58.000Z},
file_attached = {false},
profile_id = {cacab941-be62-3845-982b-a7700857a11d},
last_modified = {2016-09-07T14:54:40.000Z},
read = {false},
starred = {false},
authored = {true},
confirmed = {true},
hidden = {false},
source_type = {JOUR},
notes = {JID: 101666658; OTO: NOTNLM; 2015 [ecollection]; 2015/07/27 [received]; 2015/09/07 [accepted]; 2015/09/30 [epublish]; epublish},
bibtype = {article},
author = {McEvoy, F J},
journal = {Frontiers in veterinary science}
}Metadata Dependent Mondrian Processes. Wang, Y., Li, B., Wang, Y., & Chen, F. In Proceedings of the 32Nd International Conference on International Conference on Machine Learning - Volume 37, of ICML'15, pages 1339--1347, 2015. JMLR.org.
Paper bibtex
Paper bibtex @inproceedings{Wang:2015:MDM:3045118.3045261,
Acmid = {3045261},
Author = {Wang, Yi and Li, Bin and Wang, Yang and Chen, Fang},
Booktitle = {Proceedings of the 32Nd International Conference on International Conference on Machine Learning - Volume 37},
Date-Modified = {2017-09-27 09:00:24 +0000},
Keywords = {machine learning},
Location = {Lille, France},
Numpages = {9},
Pages = {1339--1347},
Publisher = {JMLR.org},
Series = {ICML'15},
Title = {Metadata Dependent Mondrian Processes},
Url = {http://dl.acm.org/citation.cfm?id=3045118.3045261},
Year = {2015},
Bdsk-Url-1 = {http://dl.acm.org/citation.cfm?id=3045118.3045261}}2014 (1)
ARTgrid: A Two-Level Learning Architecture Based on Adaptive Resonance Theory. Švaco, M. & Jerbić, B. Advances in Artificial Neural Systems, 2014:1–9, 2014.
doi bibtex
doi bibtex
@article{svaco_artgrid_2014,
title = {{ARTgrid}: {A} {Two}-{Level} {Learning} {Architecture} {Based} on {Adaptive} {Resonance} {Theory}},
volume = {2014},
doi = {10.1155/2014/185492},
journal = {Advances in Artificial Neural Systems},
author = {Švaco, Marko and Jerbić, Bojan},
year = {2014},
keywords = {ARTgrid, Adaptive resonance theory, Machine learning},
pages = {1--9}
}2013 (1)
An optimal control strategy with enhanced robustness for air-conditioning systems considering model and measurement uncertainties. Zhu, N., Shan, K., Wang, S., & Sun, Y. Energy and Buildings, 67:540–550, December, 2013.
Paper doi abstract bibtex
Paper doi abstract bibtex Model-based optimal controls in HVAC systems involve uncertainties due to model uncertainties and measurement uncertainties. These uncertainties affect the accuracy and reliability of the outputs of optimal control strategies, and therefore affect the energy and environmental performance of buildings. This study proposes a method to enhance the robustness of optimal control strategies. A fuzzy approach is adopted to predict the errors in models outputs. Such predicted errors are then used to correct the model outputs. The method is validated in an optimal control strategy for HVAC cooling water systems. The operation data of a real building system is used to validate the error prediction method. A simulation platform is built to validate the enhanced strategy. Measurement uncertainties are deliberately added to the simulated system for validation tests. Test results indicate that the method is effective in predicting the errors in model outputs. Significant energy savings are achieved compared with the conventional optimal control method.
@article{zhu_optimal_2013,
title = {An optimal control strategy with enhanced robustness for air-conditioning systems considering model and measurement uncertainties},
volume = {67},
issn = {0378-7788},
url = {http://www.sciencedirect.com/science/article/pii/S0378778813005471},
doi = {10.1016/j.enbuild.2013.08.050},
abstract = {Model-based optimal controls in HVAC systems involve uncertainties due to model uncertainties and measurement uncertainties. These uncertainties affect the accuracy and reliability of the outputs of optimal control strategies, and therefore affect the energy and environmental performance of buildings. This study proposes a method to enhance the robustness of optimal control strategies. A fuzzy approach is adopted to predict the errors in models outputs. Such predicted errors are then used to correct the model outputs. The method is validated in an optimal control strategy for HVAC cooling water systems. The operation data of a real building system is used to validate the error prediction method. A simulation platform is built to validate the enhanced strategy. Measurement uncertainties are deliberately added to the simulated system for validation tests. Test results indicate that the method is effective in predicting the errors in model outputs. Significant energy savings are achieved compared with the conventional optimal control method.},
journal = {Energy and Buildings},
author = {Zhu, Na and Shan, Kui and Wang, Shengwei and Sun, Yongjun},
month = dec,
year = {2013},
keywords = {Air-conditioning system, Fuzzy c-means clustering, Machine learning, Measurement uncertainty, Model uncertainty, Optimal control strategy},
pages = {540--550},
}2012 (4)
Locally-Adaptive Similarity Metric for Deformable Medical Image Registration. Tang, L. Y. W., Hero, A. O, & Hamarneh, G. In IEEE International Symposium on Biomedical Imaging (IEEE ISBI), pages 728-731, 2012.
doi bibtex
doi bibtex
@INPROCEEDINGS{isbi2012a,
OPTADDRESS = {},
AUTHOR = {Lisa Y. W. Tang and Alfred O Hero and Ghassan Hamarneh},
BOOKTITLE = {IEEE International Symposium on Biomedical Imaging (IEEE ISBI)},
OPTEDITOR = {},
OPTMONTH = {},
OPTNOTE = {},
OPTNUMBER = {},
OPTORGANIZATION = {},
PAGES = {728-731},
OPTPUBLISHER = {},
OPTSERIES = {},
TITLE = {Locally-Adaptive Similarity Metric for Deformable
Medical Image Registration},
OPTVOLUME = {},
YEAR = {2012},
OPTABSTRACT = {},
DOI = {10.1109/ISBI.2012.6235651},
OPTISBN = {},
OPTISSN = {},
KEYWORDS = {Machine Learning, Registration and Matching},
OPTURL = {},
OPTURL-PUBLISHER = {},
PDF = {http://www.cs.sfu.ca/~hamarneh/ecopy/isbi2012a.pdf}
}CrossCheck: Combining Crawling and Differencing to Better Detect Cross-browser Incompatibilities in Web Applications. Choudhary, S. R., Prasad, M. R., & Orso, A. In Proceedings of the 2012 IEEE Fifth International Conference on Software Testing, Verification and Validation, pages 171–180, Washington, DC, USA, 2012. IEEE Computer Society.
Paper doi bibtex
Paper doi bibtex @inproceedings{choudhary_crosscheck:_2012,
address = {Washington, DC, USA},
title = {{CrossCheck}: {Combining} {Crawling} and {Differencing} to {Better} {Detect} {Cross}-browser {Incompatibilities} in {Web} {Applications}},
isbn = {978-0-7695-4670-4},
url = {http://dx.doi.org/10.1109/ICST.2012.97},
doi = {10.1109/ICST.2012.97},
booktitle = {Proceedings of the 2012 {IEEE} {Fifth} {International} {Conference} on {Software} {Testing}, {Verification} and {Validation}},
publisher = {IEEE Computer Society},
author = {Choudhary, Shauvik Roy and Prasad, Mukul R. and Orso, Alessandro},
year = {2012},
keywords = {dynamic analysis, machine learning, web testing},
pages = {171--180},
}Transfer learning for cross-company software defect prediction. Ma, Y., Luo, G., Zeng, X., & Chen, A. Information and Software Technology, 54(3):248--256, 2012.
Paper doi abstract bibtex
Paper doi abstract bibtex Context Software defect prediction studies usually built models using within-company data, but very few focused on the prediction models trained with cross-company data. It is difficult to employ these models which are built on the within-company data in practice, because of the lack of these local data repositories. Recently, transfer learning has attracted more and more attention for building classifier in target domain using the data from related source domain. It is very useful in cases when distributions of training and test instances differ, but is it appropriate for cross-company software defect prediction? Objective In this paper, we consider the cross-company defect prediction scenario where source and target data are drawn from different companies. In order to harness cross company data, we try to exploit the transfer learning method to build faster and highly effective prediction model. Method Unlike the prior works selecting training data which are similar from the test data, we proposed a novel algorithm called Transfer Naive Bayes (TNB), by using the information of all the proper features in training data. Our solution estimates the distribution of the test data, and transfers cross-company data information into the weights of the training data. On these weighted data, the defect prediction model is built. Results This article presents a theoretical analysis for the comparative methods, and shows the experiment results on the data sets from different organizations. It indicates that TNB is more accurate in terms of AUC (The area under the receiver operating characteristic curve), within less runtime than the state of the art methods. Conclusion It is concluded that when there are too few local training data to train good classifiers, the useful knowledge from different-distribution training data on feature level may help. We are optimistic that our transfer learning method can guide optimal resource allocation strategies, which may reduce software testing cost and increase effectiveness of software testing process.
@article{ ma_transfer_2012,
title = {Transfer learning for cross-company software defect prediction},
volume = {54},
issn = {0950-5849},
url = {http://www.sciencedirect.com/science/article/pii/S0950584911001996},
doi = {10.1016/j.infsof.2011.09.007},
abstract = {Context
Software defect prediction studies usually built models using within-company data, but very few focused on the prediction models trained with cross-company data. It is difficult to employ these models which are built on the within-company data in practice, because of the lack of these local data repositories. Recently, transfer learning has attracted more and more attention for building classifier in target domain using the data from related source domain. It is very useful in cases when distributions of training and test instances differ, but is it appropriate for cross-company software defect prediction?
Objective
In this paper, we consider the cross-company defect prediction scenario where source and target data are drawn from different companies. In order to harness cross company data, we try to exploit the transfer learning method to build faster and highly effective prediction model.
Method
Unlike the prior works selecting training data which are similar from the test data, we proposed a novel algorithm called Transfer Naive Bayes (TNB), by using the information of all the proper features in training data. Our solution estimates the distribution of the test data, and transfers cross-company data information into the weights of the training data. On these weighted data, the defect prediction model is built.
Results
This article presents a theoretical analysis for the comparative methods, and shows the experiment results on the data sets from different organizations. It indicates that TNB is more accurate in terms of AUC (The area under the receiver operating characteristic curve), within less runtime than the state of the art methods.
Conclusion
It is concluded that when there are too few local training data to train good classifiers, the useful knowledge from different-distribution training data on feature level may help. We are optimistic that our transfer learning method can guide optimal resource allocation strategies, which may reduce software testing cost and increase effectiveness of software testing process.},
number = {3},
urldate = {2014-07-21TZ},
journal = {Information and Software Technology},
author = {Ma, Ying and Luo, Guangchun and Zeng, Xue and Chen, Aiguo},
year = {2012},
keywords = {Different distribution, Machine learning, Naive Bayes, Software defect prediction, Transfer learning, _done},
pages = {248--256}
}Triaging incoming change requests: Bug or commit history, or code authorship?. Linares-Vásquez, M., Hossen, K., Dang, H., Kagdi, H., Gethers, M., & Poshyvanyk, D. In Software Maintenance (ICSM), 2012 28th IEEE International Conference on, pages 451–460, September, 2012.
doi bibtex
doi bibtex
@inproceedings{linares-vasquez_triaging_2012,
title = {Triaging incoming change requests: {Bug} or commit history, or code authorship?},
doi = {10.1109/ICSM.2012.6405306},
booktitle = {Software {Maintenance} ({ICSM}), 2012 28th {IEEE} {International} {Conference} on},
author = {Linares-Vásquez, M. and Hossen, K. and Dang, Hoang and Kagdi, H. and Gethers, M. and Poshyvanyk, D.},
month = sep,
year = {2012},
keywords = {Accuracy, ArgoUML, Data mining, JEdit, Large scale integration, MuCommander, Software maintenance, Support vector machines, Unified modeling language, bug fixing, change request, code authorship, commit log history, expert developer recommendations, header comments, information retrieval, information retrieval technique, learning (artificial intelligence), machine learning, open source projects, program debugging, public domain software, recommendation accuracies, recommender systems, software change request textual description, source code authorship information, source code files, triaging},
pages = {451--460},
}2011 (2)
Reducing the Effort of Bug Report Triage: Recommenders for Development-oriented Decisions. Anvik, J. & Murphy, G. C. ACM Trans. Softw. Eng. Methodol., 20(3):10:1–10:35, August, 2011.
Paper doi bibtex
Paper doi bibtex @article{anvik_reducing_2011,
title = {Reducing the {Effort} of {Bug} {Report} {Triage}: {Recommenders} for {Development}-oriented {Decisions}},
volume = {20},
issn = {1049-331X},
url = {http://doi.acm.org/10.1145/2000791.2000794},
doi = {10.1145/2000791.2000794},
number = {3},
journal = {ACM Trans. Softw. Eng. Methodol.},
author = {Anvik, John and Murphy, Gail C.},
month = aug,
year = {2011},
keywords = {Bug report triage, configuration assistance, machine learning, recommendation, task assignment},
pages = {10:1--10:35},
}Online Support Vector Regression With Varying Parameters for Time-Dependent Data. Omitaomu, O. A., Jeong, M. K., & Badiru, A. B. IEEE Transactions on Systems, Man, and Cybernetics - Part A: Systems and Humans, 41(1):191–197, January, 2011. Conference Name: IEEE Transactions on Systems, Man, and Cybernetics - Part A: Systems and Humans
doi abstract bibtex
doi abstract bibtex
Support vector regression (SVR) is a machine learning technique that continues to receive interest in several domains, including manufacturing, engineering, and medicine. In order to extend its application to problems in which data sets arrive constantly and in which batch processing of the data sets is infeasible or expensive, an accurate online SVR (AOSVR) technique was proposed. The AOSVR technique efficiently updates a trained SVR function whenever a sample is added to or removed from the training set without retraining the entire training data. However, the AOSVR technique assumes that the new samples and the training samples are of the same characteristics; hence, the same value of SVR parameters is used for training and prediction. This assumption is not applicable to data samples that are inherently noisy and nonstationary, such as sensor data. As a result, we propose AOSVR with varying parameters that uses varying SVR parameters rather than fixed SVR parameters and hence accounts for the variability that may exist in the samples. To accomplish this objective, we also propose a generalized weight function to automatically update the weights of SVR parameters in online monitoring applications. The proposed function allows for lower and upper bounds for SVR parameters. We tested our proposed approach and compared results with the conventional AOSVR approach using two benchmark time-series data and sensor data from a nuclear power plant. The results show that using varying SVR parameters is more applicable to time-dependent data.
@article{omitaomu_online_2011,
title = {Online {Support} {Vector} {Regression} {With} {Varying} {Parameters} for {Time}-{Dependent} {Data}},
volume = {41},
issn = {1558-2426},
doi = {10.1109/TSMCA.2010.2055156},
abstract = {Support vector regression (SVR) is a machine learning technique that continues to receive interest in several domains, including manufacturing, engineering, and medicine. In order to extend its application to problems in which data sets arrive constantly and in which batch processing of the data sets is infeasible or expensive, an accurate online SVR (AOSVR) technique was proposed. The AOSVR technique efficiently updates a trained SVR function whenever a sample is added to or removed from the training set without retraining the entire training data. However, the AOSVR technique assumes that the new samples and the training samples are of the same characteristics; hence, the same value of SVR parameters is used for training and prediction. This assumption is not applicable to data samples that are inherently noisy and nonstationary, such as sensor data. As a result, we propose AOSVR with varying parameters that uses varying SVR parameters rather than fixed SVR parameters and hence accounts for the variability that may exist in the samples. To accomplish this objective, we also propose a generalized weight function to automatically update the weights of SVR parameters in online monitoring applications. The proposed function allows for lower and upper bounds for SVR parameters. We tested our proposed approach and compared results with the conventional AOSVR approach using two benchmark time-series data and sensor data from a nuclear power plant. The results show that using varying SVR parameters is more applicable to time-dependent data.},
number = {1},
journal = {IEEE Transactions on Systems, Man, and Cybernetics - Part A: Systems and Humans},
author = {Omitaomu, Olufemi A. and Jeong, Myong K. and Badiru, Adedeji B.},
month = jan,
year = {2011},
note = {Conference Name: IEEE Transactions on Systems, Man, and Cybernetics - Part A: Systems and Humans},
keywords = {Automobile manufacture, Condition monitoring, Data engineering, Machine learning, Manufacturing, Medical diagnostic imaging, Power generation, Sensor systems, Systems engineering and theory, Training data, inferential sensing, online prediction, support vector machine, system diagnosis},
pages = {191--197},
}2010 (1)
Lexicons and Grammars for Named Entity Annotation in the National Corpus of Polish. Savary, A. & Piskorski, J. Information Systems Journal, 2010.
Website abstract bibtex
Website abstract bibtex The paper investigates the accuracy of a Named Entity Recognition (NER) algorithm based on the Hidden Markov Model in the domain of Polish stock exchange reports. The task of NER was limited to the recognition and classification of Named Entities representing persons and companies. The algorithm was tested on a small Polish domain corpus of stock exchange reports. A comparison with the baselines of the algorithms based on the case of the first letters and a gazetteer is presented. The algorithm outperformed both baselines; it achieved 64% precision and 93% recall for person names and 78% precision and 83% recall for company names. Introduction of simple hand-written post-processing rules increased the precision for person names up to 87%. A cross-domain evaluation on a small corpus of police reports is also presented. We discuss the problem of method portability in relation to much worse results obtained on the second corpus. A possible combination of different knowledge sources is sketched as a possible way of overcoming the portability problem.
@article{
title = {Lexicons and Grammars for Named Entity Annotation in the National Corpus of Polish},
type = {article},
year = {2010},
identifiers = {[object Object]},
keywords = {Hidden Markov Model,Machine Learning,Named Entity Recognition},
pages = {141-154},
websites = {http://iis.ipipan.waw.pl/2010/proceedings/iis10-14.pdf},
editors = {[object Object],[object Object],[object Object],[object Object],[object Object]},
id = {e122c9c8-c964-30c5-9b30-ab44b1819775},
created = {2012-01-21T12:35:31.000Z},
file_attached = {false},
profile_id = {5284e6aa-156c-3ce5-bc0e-b80cf09f3ef6},
group_id = {066b42c8-f712-3fc3-abb2-225c158d2704},
last_modified = {2017-03-14T14:36:19.698Z},
tags = {named entity recognition},
read = {false},
starred = {false},
authored = {false},
confirmed = {true},
hidden = {false},
citation_key = {Savary2010},
private_publication = {false},
abstract = {The paper investigates the accuracy of a Named Entity Recognition (NER) algorithm based on the Hidden Markov Model in the domain of Polish stock exchange reports. The task of NER was limited to the recognition and classification of Named Entities representing persons and companies. The algorithm was tested on a small Polish domain corpus of stock exchange reports. A comparison with the baselines of the algorithms based on the case of the first letters and a gazetteer is presented. The algorithm outperformed both baselines; it achieved 64% precision and 93% recall for person names and 78% precision and 83% recall for company names. Introduction of simple hand-written post-processing rules increased the precision for person names up to 87%. A cross-domain evaluation on a small corpus of police reports is also presented. We discuss the problem of method portability in relation to much worse results obtained on the second corpus. A possible combination of different knowledge sources is sketched as a possible way of overcoming the portability problem.},
bibtype = {article},
author = {Savary, Agata and Piskorski, Jakub},
journal = {Information Systems Journal}
}2008 (2)
3D Bicipital Groove Shape Analysis and Relationship to Tendopathy. Ward, A., Hamarneh, G., & Schweitzer, M. Journal of Digital Imaging, 21(2):219-234, 2008.
doi bibtex
doi bibtex
@ARTICLE{jdi2008a,
AUTHOR = {Aaron Ward and Ghassan Hamarneh and Mark Schweitzer},
JOURNAL = {Journal of Digital Imaging},
OPTMONTH = {},
OPTNOTE = {},
NUMBER = {2},
PAGES = {219-234},
TITLE = {3D Bicipital Groove Shape Analysis and Relationship to
Tendopathy},
VOLUME = {21},
YEAR = {2008},
OPTABSTRACT = {},
DOI = {10.1007/s10278-007-9027-6},
OPTISBN = {},
OPTISSN = {},
KEYWORDS = {Shape Modelling and Analysis, Machine Learning},
OPTURL = {},
OPTURL-PUBLISHER = {},
PDF = {http://www.cs.sfu.ca/~hamarneh/ecopy/jdi2008a.pdf}
}Comparison of classification accuracy using Cohen’s Weighted Kappa. Ben-David, A. Expert Systems with Applications, 34(2):825–832, February, 2008.
Paper doi abstract bibtex
Paper doi abstract bibtex Many expert systems solve classification problems. While comparing the accuracy of such classifiers, the cost of error must frequently be taken into account. In such cost-sensitive applications just using the percentage of misses as the sole meter for accuracy can be misleading. Typical examples of such problems are medical and military applications, as well as data sets with ordinal (i.e., ordered) class. A new methodology is proposed here for assessing classifiers accuracy. The approach taken is based on Cohen’s Kappa statistic. It compensates for classifications that may be due to chance. The use of Kappa is proposed as a standard meter for measuring the accuracy of all multi-valued classification problems. The use of Weighted Kappa enables to effectively deal with cost-sensitive classification. When the cost of error is unknown and can only be roughly estimated, the use of sensitivity analysis with Weighted Kappa is highly recommended.
@article{ben-david_comparison_2008,
title = {Comparison of classification accuracy using {Cohen}’s {Weighted} {Kappa}},
volume = {34},
issn = {0957-4174},
url = {http://www.sciencedirect.com/science/article/pii/S0957417406003435},
doi = {10.1016/j.eswa.2006.10.022},
abstract = {Many expert systems solve classification problems. While comparing the accuracy of such classifiers, the cost of error must frequently be taken into account. In such cost-sensitive applications just using the percentage of misses as the sole meter for accuracy can be misleading. Typical examples of such problems are medical and military applications, as well as data sets with ordinal (i.e., ordered) class.
A new methodology is proposed here for assessing classifiers accuracy. The approach taken is based on Cohen’s Kappa statistic. It compensates for classifications that may be due to chance. The use of Kappa is proposed as a standard meter for measuring the accuracy of all multi-valued classification problems. The use of Weighted Kappa enables to effectively deal with cost-sensitive classification. When the cost of error is unknown and can only be roughly estimated, the use of sensitivity analysis with Weighted Kappa is highly recommended.},
number = {2},
urldate = {2014-11-09},
journal = {Expert Systems with Applications},
author = {Ben-David, Arie},
month = feb,
year = {2008},
keywords = {Cost-sensitive classification, Expert systems, Machine learning, Ordinal data sets, Sensitivity analysis, Weighted Cohen’s Kappa},
pages = {825--832},
}2006 (1)
Annotation guidelines for machine learning-based named entity recognition in microbiology. Nédellec, C., Bessieres, P., Bossy, R., Kptoujanksy, A., & Manine, A., P. Machine Learning, Citeseer, 2006.
Website abstract bibtex
Website abstract bibtex Recent challenges on machine learning application to named-entity recognition in biology trigger discussions on the manual annotation guidelines for annotating the learning corpora. Some sources of potential inconsistency have been identified by corpus annotators and challenge participants. We go one step further by proposing specific annotation guidelines for biology and evaluating their effect on performances of machine learning methods. We show that a significant improvement can be achieved by this way that is not due to the feature set neither to the ML methods.
@article{
title = {Annotation guidelines for machine learning-based named entity recognition in microbiology},
type = {article},
year = {2006},
keywords = {annotation guidelines,machine learning,named entity recognition},
pages = {40–54},
websites = {http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.112.6401&rep=rep1&type=pdf},
publisher = {Citeseer},
id = {4e2e0ad6-9fef-399c-8707-68f2c6b464f3},
created = {2011-12-28T07:04:55.000Z},
file_attached = {false},
profile_id = {5284e6aa-156c-3ce5-bc0e-b80cf09f3ef6},
group_id = {066b42c8-f712-3fc3-abb2-225c158d2704},
last_modified = {2017-03-14T14:36:19.698Z},
tags = {named entities},
read = {false},
starred = {false},
authored = {false},
confirmed = {true},
hidden = {false},
citation_key = {Nedellec2006},
private_publication = {false},
abstract = {Recent challenges on machine learning application to named-entity recognition in biology trigger discussions on the manual annotation guidelines for annotating the learning corpora. Some sources of potential inconsistency have been identified by corpus annotators and challenge participants. We go one step further by proposing specific annotation guidelines for biology and evaluating their effect on performances of machine learning methods. We show that a significant improvement can be achieved by this way that is not due to the feature set neither to the ML methods.},
bibtype = {article},
author = {Nédellec, C and Bessieres, P and Bossy, Robert and Kptoujanksy, A and Manine, A P},
journal = {Machine Learning}
}2001 (1)
Variable Selection via Nonconcave Penalized Likelihood and its Oracle Properties. Fan, J. & Li, R. Journal of the American Statistical Association, 96(456):1348–1360, December, 2001.
Paper doi bibtex
Paper doi bibtex @article{fan_variable_2001,
title = {Variable {Selection} via {Nonconcave} {Penalized} {Likelihood} and its {Oracle} {Properties}},
volume = {96},
issn = {0162-1459, 1537-274X},
url = {http://www.tandfonline.com/doi/abs/10.1198/016214501753382273},
doi = {10.1198/016214501753382273},
language = {en},
number = {456},
urldate = {2021-07-06},
journal = {Journal of the American Statistical Association},
author = {Fan, Jianqing and Li, Runze},
month = dec,
year = {2001},
keywords = {Lasso, ML, Machine Learning},
pages = {1348--1360},
}2000 (1)
Towards Learning New Methods in Proof Planning. Jamnik, M., Kerber, M., & Benzmüller, C. In Kerber, M. & Kohlhase, M., editors, Symbolic Computation and Automated Reasoning, pages 142-159, 2000. A.K.Peters.
Preprint abstract bibtex
Preprint abstract bibtex In this paper we propose how proof planning systems can be extended by an automated learning capability. The idea is that a proof planner would be capable of learning new proof methods from well chosen examples of proofs which use a similar reasoning strategy to prove related theorems, and this strategy could be characterised as a proof method. We propose a representation framework for methods, and a machine learning technique which can learn methods using this representation framework. The technique can be applied (amongst other) to learn whether and when to call external systems such as theorem provers or computer algebra systems. This is work in progress, and we hope to gain useful feedback from the CALCULEMUS community.
@inproceedings{C9,
Abstract = {In this paper we propose how proof planning systems
can be extended by an automated learning
capability. The idea is that a proof planner would
be capable of learning new proof methods from well
chosen examples of proofs which use a similar
reasoning strategy to prove related theorems, and
this strategy could be characterised as a proof
method. We propose a representation framework for
methods, and a machine learning technique which can
learn methods using this representation
framework. The technique can be applied (amongst
other) to learn whether and when to call external
systems such as theorem provers or computer algebra
systems. This is work in progress, and we hope to
gain useful feedback from the CALCULEMUS community.},
Author = {Mateja Jamnik and Manfred Kerber and Christoph
Benzm{\"u}ller},
Booktitle = {Symbolic Computation and Automated Reasoning},
Editor = {Manfred Kerber and Michael Kohlhase},
Keywords = {own, CALCULEMUS, Symb. Computation and
Symb. Reasoning, Proof Planning, Machine Learning},
Pages = {142-159},
Publisher = {A.K.Peters},
Title = {Towards Learning New Methods in Proof Planning},
url_preprint = {http://christoph-benzmueller.de/papers/C9.pdf},
Year = 2000,
}1999 (1)
An overview of statistical learning theory. Vapnik, V. N. IEEE Trans Neural Netw, 10(5):988–999, 1999.
doi bibtex
doi bibtex
@Article{vapnik99overview,
author = {V. N. Vapnik},
title = {An overview of statistical learning theory},
journal = {IEEE Trans Neural Netw},
year = {1999},
volume = {10},
number = {5},
pages = {988--999},
optmonth = sep,
issn = {1045-9227},
doi = {10.1109/72.788640},
keywords = {estimation theory, generalisation (artificial intelligence), learning (artificial intelligence), statistical analysis, function estimation, generalization conditions, multidimensional function estimation, statistical learning theory, support vector machines, Algorithm design and analysis, Loss measurement, Machine learning, Multidimensional systems, Pattern recognition, Probability distribution, Risk management, Statistical learning, Support vector machines},
}1994 (1)
On a learnability question associated to neural networks with continuous activations (extended abstract). DasGupta, B., Siegelmann, H. T., & Sontag, E. In COLT '94: Proceedings of the seventh annual conference on Computational learning theory, pages 47–56, New York, NY, USA, 1994. ACM Press.
doi bibtex
doi bibtex
@INPROCEEDINGS{181009,
AUTHOR = {B. DasGupta and H. T. Siegelmann and E.D. Sontag},
BOOKTITLE = {COLT '94: Proceedings of the seventh annual conference on Computational learning theory},
TITLE = {On a learnability question associated to neural networks
with continuous activations (extended abstract)},
YEAR = {1994},
ADDRESS = {New York, NY, USA},
OPTCROSSREF = {},
OPTEDITOR = {},
OPTMONTH = {},
OPTNOTE = {},
OPTNUMBER = {},
OPTORGANIZATION = {},
PAGES = {47--56},
PUBLISHER = {ACM Press},
OPTSERIES = {},
OPTVOLUME = {},
KEYWORDS = {analog computing, neural networks,
computational complexity, machine learning},
DOI = {http://doi.acm.org/10.1145/180139.181009}
}undefined (2)
Thoughtful Machine Learning with Python: A Test-Driven Approach. Kirk, M. O'Reilly, First edition edition. OCLC: ocn908375399
bibtex
bibtex
@book{kirkThoughtfulMachineLearning2017,
location = {{Beijing ; Boston}},
title = {Thoughtful Machine Learning with {{Python}}: A Test-Driven Approach},
edition = {First edition},
isbn = {978-1-4919-2413-6},
shorttitle = {Thoughtful Machine Learning with {{Python}}},
pagetotal = {201},
publisher = {{O'Reilly}},
date = {2017},
keywords = {Machine learning,Python (Computer program language)},
author = {Kirk, Matthew},
file = {/home/dimitri/Nextcloud/Zotero/storage/CSDF3XWI/thoughtfulmachinelearningwithpython.pdf},
note = {OCLC: ocn908375399}
}Exploring Reinforcement Learning for Mobile Percussive Collaboration. Derbinsky, N. & Essl, G. Technical Report
abstract bibtex
abstract bibtex
This paper presents a system for mobile percussive collaboration. We show that reinforcement learning can incremen-tally learn percussive beat patterns played by humans and supports real-time collaborative performance in the absence of one or more performers. This work leverages an existing integration between urMus and Soar and addresses multiple challenges involved in the deployment of machine-learning algorithms for mobile music expression, including tradeoffs between learning speed & quality; interface design for human collaborators; and real-time performance and improvisation .
@techreport{derbinsky_exploring_nodate,
title = {Exploring {Reinforcement} {Learning} for {Mobile} {Percussive} {Collaboration}},
abstract = {This paper presents a system for mobile percussive collaboration. We show that reinforcement learning can incremen-tally learn percussive beat patterns played by humans and supports real-time collaborative performance in the absence of one or more performers. This work leverages an existing integration between urMus and Soar and addresses multiple challenges involved in the deployment of machine-learning algorithms for mobile music expression, including tradeoffs between learning speed \& quality; interface design for human collaborators; and real-time performance and improvisation .},
urldate = {2019-09-04},
author = {Derbinsky, Nate and Essl, Georg},
keywords = {Mobile music, cognitive architecture, machine learning},
}