{"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/mobile-sensor-data-anonymization","title":"Mobile Sensor Data Anonymization","arxiv_id":"1810.11546","date":"2018-10-26","proceeding":null,"authors":["Mohammad Malekzadeh","Richard G. Clegg","Andrea Cavallaro","Hamed Haddadi"],"abstract":"Motion sensors such as accelerometers and gyroscopes measure the instant\nacceleration and rotation of a device, in three dimensions. Raw data streams\nfrom motion sensors embedded in portable and wearable devices may reveal\nprivate information about users without their awareness. For example, motion\ndata might disclose the weight or gender of a user, or enable their\nre-identification. To address this problem, we propose an on-device\ntransformation of sensor data to be shared for specific applications, such as\nmonitoring selected daily activities, without revealing information that\nenables user identification. We formulate the anonymization problem using an\ninformation-theoretic approach and propose a new multi-objective loss function\nfor training deep autoencoders. This loss function helps minimizing\nuser-identity information as well as data distortion to preserve the\napplication-specific utility. The training process regulates the encoder to\ndisregard user-identifiable patterns and tunes the decoder to shape the output\nindependently of users in the training set. The trained autoencoder can be\ndeployed on a mobile or wearable device to anonymize sensor data even for users\nwho are not included in the training dataset. Data from 24 users transformed by\nthe proposed anonymizing autoencoder lead to a promising trade-off between\nutility and privacy, with an accuracy for activity recognition above 92% and an\naccuracy for user identification below 7%.","url_abs":"http://arxiv.org/abs/1810.11546v3","url_pdf":"http://arxiv.org/pdf/1810.11546v3.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":"mobile-sensor-data-anonymization","repo_url":"https://github.com/mmalekzadeh/motion-sense","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[{"task_slug":"activity-recognition","task_name":"Activity Recognition"},{"task_slug":"decoder","task_name":"Decoder"},{"task_slug":"user-identification","task_name":"User Identification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1810.11546","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}