Deep Multimodal Data Fusion. Zhao, F., Zhang, C., & Geng, B. ACM Computing Surveys, 56(9):1–36, October, 2024. ![Deep Multimodal Data Fusion [link]](https://bibbase.org/img/filetypes/link.svg "link")
Paper doi abstract bibtex Multimodal Artificial Intelligence (Multimodal AI), in general, involves various types of data (e.g., images, texts, or data collected from different sensors), feature engineering (e.g., extraction, combination/fusion), and decision-making (e.g., majority vote). As architectures become more and more sophisticated, multimodal neural networks can integrate feature extraction, feature fusion, and decision-making processes into one single model. The boundaries between those processes are increasingly blurred. The conventional multimodal data fusion taxonomy (e.g., early/late fusion), based on which the fusion occurs in, is no longer suitable for the modern deep learning era. Therefore, based on the main-stream techniques used, we propose a new fine-grained taxonomy grouping the state-of-the-art (SOTA) models into five classes: Encoder-Decoder methods, Attention Mechanism methods, Graph Neural Network methods, Generative Neural Network methods, and other Constraint-based methods. Most existing surveys on multimodal data fusion are only focused on one specific task with a combination of two specific modalities. Unlike those, this survey covers a broader combination of modalities, including Vision + Language (e.g., videos, texts), Vision + Sensors (e.g., images, LiDAR), and so on, and their corresponding tasks (e.g., video captioning, object detection). Moreover, a comparison among these methods is provided, as well as challenges and future directions in this area.
@article{zhao_deep_2024,
title = {Deep {Multimodal} {Data} {Fusion}},
volume = {56},
issn = {0360-0300, 1557-7341},
url = {https://dl.acm.org/doi/10.1145/3649447},
doi = {10.1145/3649447},
abstract = {Multimodal Artificial Intelligence (Multimodal AI), in general, involves various types of data (e.g., images, texts, or data collected from different sensors), feature engineering (e.g., extraction, combination/fusion), and decision-making (e.g., majority vote). As architectures become more and more sophisticated, multimodal neural networks can integrate feature extraction, feature fusion, and decision-making processes into one single model. The boundaries between those processes are increasingly blurred. The conventional multimodal data fusion taxonomy (e.g., early/late fusion), based on which the fusion occurs in, is no longer suitable for the modern deep learning era. Therefore, based on the main-stream techniques used, we propose a new fine-grained taxonomy grouping the state-of-the-art (SOTA) models into five classes: Encoder-Decoder methods, Attention Mechanism methods, Graph Neural Network methods, Generative Neural Network methods, and other Constraint-based methods. Most existing surveys on multimodal data fusion are only focused on one specific task with a combination of two specific modalities. Unlike those, this survey covers a broader combination of modalities, including Vision + Language (e.g., videos, texts), Vision + Sensors (e.g., images, LiDAR), and so on, and their corresponding tasks (e.g., video captioning, object detection). Moreover, a comparison among these methods is provided, as well as challenges and future directions in this area.},
language = {en},
number = {9},
urldate = {2025-03-09},
journal = {ACM Computing Surveys},
author = {Zhao, Fei and Zhang, Chengcui and Geng, Baocheng},
month = oct,
year = {2024},
pages = {1--36},
}
Downloads: 0
{"_id":"uc2k8Lud3nynkEyjm","bibbaseid":"zhao-zhang-geng-deepmultimodaldatafusion-2024","author_short":["Zhao, F.","Zhang, C.","Geng, B."],"bibdata":{"bibtype":"article","type":"article","title":"Deep Multimodal Data Fusion","volume":"56","issn":"0360-0300, 1557-7341","url":"https://dl.acm.org/doi/10.1145/3649447","doi":"10.1145/3649447","abstract":"Multimodal Artificial Intelligence (Multimodal AI), in general, involves various types of data (e.g., images, texts, or data collected from different sensors), feature engineering (e.g., extraction, combination/fusion), and decision-making (e.g., majority vote). As architectures become more and more sophisticated, multimodal neural networks can integrate feature extraction, feature fusion, and decision-making processes into one single model. The boundaries between those processes are increasingly blurred. The conventional multimodal data fusion taxonomy (e.g., early/late fusion), based on which the fusion occurs in, is no longer suitable for the modern deep learning era. Therefore, based on the main-stream techniques used, we propose a new fine-grained taxonomy grouping the state-of-the-art (SOTA) models into five classes: Encoder-Decoder methods, Attention Mechanism methods, Graph Neural Network methods, Generative Neural Network methods, and other Constraint-based methods. Most existing surveys on multimodal data fusion are only focused on one specific task with a combination of two specific modalities. Unlike those, this survey covers a broader combination of modalities, including Vision + Language (e.g., videos, texts), Vision + Sensors (e.g., images, LiDAR), and so on, and their corresponding tasks (e.g., video captioning, object detection). Moreover, a comparison among these methods is provided, as well as challenges and future directions in this area.","language":"en","number":"9","urldate":"2025-03-09","journal":"ACM Computing Surveys","author":[{"propositions":[],"lastnames":["Zhao"],"firstnames":["Fei"],"suffixes":[]},{"propositions":[],"lastnames":["Zhang"],"firstnames":["Chengcui"],"suffixes":[]},{"propositions":[],"lastnames":["Geng"],"firstnames":["Baocheng"],"suffixes":[]}],"month":"October","year":"2024","pages":"1–36","bibtex":"@article{zhao_deep_2024,\n\ttitle = {Deep {Multimodal} {Data} {Fusion}},\n\tvolume = {56},\n\tissn = {0360-0300, 1557-7341},\n\turl = {https://dl.acm.org/doi/10.1145/3649447},\n\tdoi = {10.1145/3649447},\n\tabstract = {Multimodal Artificial Intelligence (Multimodal AI), in general, involves various types of data (e.g., images, texts, or data collected from different sensors), feature engineering (e.g., extraction, combination/fusion), and decision-making (e.g., majority vote). As architectures become more and more sophisticated, multimodal neural networks can integrate feature extraction, feature fusion, and decision-making processes into one single model. The boundaries between those processes are increasingly blurred. The conventional multimodal data fusion taxonomy (e.g., early/late fusion), based on which the fusion occurs in, is no longer suitable for the modern deep learning era. Therefore, based on the main-stream techniques used, we propose a new fine-grained taxonomy grouping the state-of-the-art (SOTA) models into five classes: Encoder-Decoder methods, Attention Mechanism methods, Graph Neural Network methods, Generative Neural Network methods, and other Constraint-based methods. Most existing surveys on multimodal data fusion are only focused on one specific task with a combination of two specific modalities. Unlike those, this survey covers a broader combination of modalities, including Vision + Language (e.g., videos, texts), Vision + Sensors (e.g., images, LiDAR), and so on, and their corresponding tasks (e.g., video captioning, object detection). Moreover, a comparison among these methods is provided, as well as challenges and future directions in this area.},\n\tlanguage = {en},\n\tnumber = {9},\n\turldate = {2025-03-09},\n\tjournal = {ACM Computing Surveys},\n\tauthor = {Zhao, Fei and Zhang, Chengcui and Geng, Baocheng},\n\tmonth = oct,\n\tyear = {2024},\n\tpages = {1--36},\n}\n\n\n\n\n\n\n\n","author_short":["Zhao, F.","Zhang, C.","Geng, B."],"key":"zhao_deep_2024","id":"zhao_deep_2024","bibbaseid":"zhao-zhang-geng-deepmultimodaldatafusion-2024","role":"author","urls":{"Paper":"https://dl.acm.org/doi/10.1145/3649447"},"metadata":{"authorlinks":{}},"downloads":0,"html":""},"bibtype":"article","biburl":"https://bibbase.org/zotero/yywangvr","dataSources":["PW3eQRZmFcK6vLuar"],"keywords":[],"search_terms":["deep","multimodal","data","fusion","zhao","zhang","geng"],"title":"Deep Multimodal Data Fusion","year":2024}