Unsupervised Diffusion Model for Sensor-based Human Activity Recognition. Zuo, S., Fortes, V., Suh, S., Sigg, S., & Lukowicz, P. In Adjunct Proceedings of the 2023 ACM International Joint Conference on Pervasive and Ubiquitous Computing & the 2023 ACM International Symposium on Wearable Computing, of UbiComp/ISWC '23 Adjunct, pages 205, New York, NY, USA, 2023. Association for Computing Machinery. ![Unsupervised Diffusion Model for Sensor-based Human Activity Recognition [link]](https://bibbase.org/img/filetypes/link.svg "link")
Paper doi abstract bibtex Recognizing human activities from sensor data is a vital task in various domains, but obtaining diverse and labeled sensor data remains challenging and costly. In this paper, we propose an unsupervised statistical feature-guided diffusion model for sensor-based human activity recognition. The proposed method aims to generate synthetic time-series sensor data without relying on labeled data, addressing the scarcity and annotation difficulties associated with real-world sensor data. By conditioning the diffusion model on statistical information such as mean, standard deviation, Z-score, and skewness, we generate diverse and representative synthetic sensor data. We conducted experiments on public human activity recognition datasets and compared the proposed method to conventional oversampling methods and state-of-the-art generative adversarial network methods. The experimental results demonstrate that the proposed method can improve the performance of human activity recognition and outperform existing techniques.
@inproceedings{10.1145/3594739.3610797,
author = {Zuo, Si and Fortes, Vitor and Suh, Sungho and Sigg, Stephan and Lukowicz, Paul},
title = {Unsupervised Diffusion Model for Sensor-based Human Activity Recognition},
year = {2023},
isbn = {9798400702006},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
url = {https://doi.org/10.1145/3594739.3610797},
doi = {10.1145/3594739.3610797},
abstract = {Recognizing human activities from sensor data is a vital task in various domains, but obtaining diverse and labeled sensor data remains challenging and costly. In this paper, we propose an unsupervised statistical feature-guided diffusion model for sensor-based human activity recognition. The proposed method aims to generate synthetic time-series sensor data without relying on labeled data, addressing the scarcity and annotation difficulties associated with real-world sensor data. By conditioning the diffusion model on statistical information such as mean, standard deviation, Z-score, and skewness, we generate diverse and representative synthetic sensor data. We conducted experiments on public human activity recognition datasets and compared the proposed method to conventional oversampling methods and state-of-the-art generative adversarial network methods. The experimental results demonstrate that the proposed method can improve the performance of human activity recognition and outperform existing techniques.},
booktitle = {Adjunct Proceedings of the 2023 ACM International Joint Conference on Pervasive and Ubiquitous Computing \& the 2023 ACM International Symposium on Wearable Computing},
pages = {205},
numpages = {1},
keywords = {Human activity recognition, Sensor data generation, Statistical feature-guided diffusion model, Unsupervised learning},
location = {Cancun, Quintana Roo, Mexico},
series = {UbiComp/ISWC '23 Adjunct}
}
Downloads: 0
{"_id":"bjAht7nNFc82TjSbQ","bibbaseid":"zuo-fortes-suh-sigg-lukowicz-unsuperviseddiffusionmodelforsensorbasedhumanactivityrecognition-2023","author_short":["Zuo, S.","Fortes, V.","Suh, S.","Sigg, S.","Lukowicz, P."],"bibdata":{"bibtype":"inproceedings","type":"inproceedings","author":[{"propositions":[],"lastnames":["Zuo"],"firstnames":["Si"],"suffixes":[]},{"propositions":[],"lastnames":["Fortes"],"firstnames":["Vitor"],"suffixes":[]},{"propositions":[],"lastnames":["Suh"],"firstnames":["Sungho"],"suffixes":[]},{"propositions":[],"lastnames":["Sigg"],"firstnames":["Stephan"],"suffixes":[]},{"propositions":[],"lastnames":["Lukowicz"],"firstnames":["Paul"],"suffixes":[]}],"title":"Unsupervised Diffusion Model for Sensor-based Human Activity Recognition","year":"2023","isbn":"9798400702006","publisher":"Association for Computing Machinery","address":"New York, NY, USA","url":"https://doi.org/10.1145/3594739.3610797","doi":"10.1145/3594739.3610797","abstract":"Recognizing human activities from sensor data is a vital task in various domains, but obtaining diverse and labeled sensor data remains challenging and costly. In this paper, we propose an unsupervised statistical feature-guided diffusion model for sensor-based human activity recognition. The proposed method aims to generate synthetic time-series sensor data without relying on labeled data, addressing the scarcity and annotation difficulties associated with real-world sensor data. By conditioning the diffusion model on statistical information such as mean, standard deviation, Z-score, and skewness, we generate diverse and representative synthetic sensor data. We conducted experiments on public human activity recognition datasets and compared the proposed method to conventional oversampling methods and state-of-the-art generative adversarial network methods. The experimental results demonstrate that the proposed method can improve the performance of human activity recognition and outperform existing techniques.","booktitle":"Adjunct Proceedings of the 2023 ACM International Joint Conference on Pervasive and Ubiquitous Computing & the 2023 ACM International Symposium on Wearable Computing","pages":"205","numpages":"1","keywords":"Human activity recognition, Sensor data generation, Statistical feature-guided diffusion model, Unsupervised learning","location":"Cancun, Quintana Roo, Mexico","series":"UbiComp/ISWC '23 Adjunct","bibtex":"@inproceedings{10.1145/3594739.3610797,\n author = {Zuo, Si and Fortes, Vitor and Suh, Sungho and Sigg, Stephan and Lukowicz, Paul},\n title = {Unsupervised Diffusion Model for Sensor-based Human Activity Recognition},\n year = {2023},\n isbn = {9798400702006},\n publisher = {Association for Computing Machinery},\n address = {New York, NY, USA},\n url = {https://doi.org/10.1145/3594739.3610797},\n doi = {10.1145/3594739.3610797},\n abstract = {Recognizing human activities from sensor data is a vital task in various domains, but obtaining diverse and labeled sensor data remains challenging and costly. In this paper, we propose an unsupervised statistical feature-guided diffusion model for sensor-based human activity recognition. The proposed method aims to generate synthetic time-series sensor data without relying on labeled data, addressing the scarcity and annotation difficulties associated with real-world sensor data. By conditioning the diffusion model on statistical information such as mean, standard deviation, Z-score, and skewness, we generate diverse and representative synthetic sensor data. We conducted experiments on public human activity recognition datasets and compared the proposed method to conventional oversampling methods and state-of-the-art generative adversarial network methods. The experimental results demonstrate that the proposed method can improve the performance of human activity recognition and outperform existing techniques.},\n booktitle = {Adjunct Proceedings of the 2023 ACM International Joint Conference on Pervasive and Ubiquitous Computing \\& the 2023 ACM International Symposium on Wearable Computing},\n pages = {205},\n numpages = {1},\n keywords = {Human activity recognition, Sensor data generation, Statistical feature-guided diffusion model, Unsupervised learning},\n location = {Cancun, Quintana Roo, Mexico},\n series = {UbiComp/ISWC '23 Adjunct}\n }\n\n","author_short":["Zuo, S.","Fortes, V.","Suh, S.","Sigg, S.","Lukowicz, P."],"key":"10.1145/3594739.3610797","id":"10.1145/3594739.3610797","bibbaseid":"zuo-fortes-suh-sigg-lukowicz-unsuperviseddiffusionmodelforsensorbasedhumanactivityrecognition-2023","role":"author","urls":{"Paper":"https://doi.org/10.1145/3594739.3610797"},"keyword":["Human activity recognition","Sensor data generation","Statistical feature-guided diffusion model","Unsupervised learning"],"metadata":{"authorlinks":{}}},"bibtype":"inproceedings","biburl":"http://ambientintelligence.aalto.fi/bibtex/LiteraturAll","dataSources":["aPfcTvMp5jE2KuS7H","a6QYyvmdLfrsx7DiL"],"keywords":["human activity recognition","sensor data generation","statistical feature-guided diffusion model","unsupervised learning"],"search_terms":["unsupervised","diffusion","model","sensor","based","human","activity","recognition","zuo","fortes","suh","sigg","lukowicz"],"title":"Unsupervised Diffusion Model for Sensor-based Human Activity Recognition","year":2023,"downloads":1}