{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/deepsense-a-unified-deep-learning-framework","title":"DeepSense: A Unified Deep Learning Framework for Time-Series Mobile Sensing Data Processing","arxiv_id":"1611.01942","date":"2016-11-07","proceeding":null,"authors":["Shuochao Yao","Shaohan Hu","Yiran Zhao","Aston Zhang","Tarek Abdelzaher"],"abstract":"Mobile sensing applications usually require time-series inputs from sensors.\nSome applications, such as tracking, can use sensed acceleration and rate of\nrotation to calculate displacement based on physical system models. Other\napplications, such as activity recognition, extract manually designed features\nfrom sensor inputs for classification. Such applications face two challenges.\nOn one hand, on-device sensor measurements are noisy. For many mobile\napplications, it is hard to find a distribution that exactly describes the\nnoise in practice. Unfortunately, calculating target quantities based on\nphysical system and noise models is only as accurate as the noise assumptions.\nSimilarly, in classification applications, although manually designed features\nhave proven to be effective, it is not always straightforward to find the most\nrobust features to accommodate diverse sensor noise patterns and user\nbehaviors. To this end, we propose DeepSense, a deep learning framework that\ndirectly addresses the aforementioned noise and feature customization\nchallenges in a unified manner. DeepSense integrates convolutional and\nrecurrent neural networks to exploit local interactions among similar mobile\nsensors, merge local interactions of different sensory modalities into global\ninteractions, and extract temporal relationships to model signal dynamics.\nDeepSense thus provides a general signal estimation and classification\nframework that accommodates a wide range of applications. We demonstrate the\neffectiveness of DeepSense using three representative and challenging tasks:\ncar tracking with motion sensors, heterogeneous human activity recognition, and\nuser identification with biometric motion analysis. DeepSense significantly\noutperforms the state-of-the-art methods for all three tasks. In addition,\nDeepSense is feasible to implement on smartphones due to its moderate energy\nconsumption and low latency","url_abs":"http://arxiv.org/abs/1611.01942v2","url_pdf":"http://arxiv.org/pdf/1611.01942v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"deepsense-a-unified-deep-learning-framework","repo_url":"https://github.com/DigitalBiomarkerDiscoveryPipeline/Human-Activity-Recognition","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"activity-recognition","task_name":"Activity Recognition"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"human-activity-recognition","task_name":"Human Activity Recognition"},{"task_slug":"question-answering","task_name":"Question Answering"},{"task_slug":"time-series-1","task_name":"Time Series"},{"task_slug":"time-series","task_name":"Time Series Analysis"},{"task_slug":"user-identification","task_name":"User Identification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/question-answering-on-newsqa","task":"Question Answering","dataset":"NewsQA","model":"Riple/Saanvi-v0.5-DeepAnalysis","rank_in_archive_order":2,"of":18,"metrics":{"EM":"92.14","F1":"94.01"},"uses_additional_data":true}],"syntology":{"syntology_url":"https://syntology.ai/paper/1611.01942","atlas_url":"https://app.syntology.ai/?focus=1611.01942","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}