{"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/preserving-differential-privacy-in","title":"Preserving Differential Privacy in Convolutional Deep Belief Networks","arxiv_id":"1706.08839","date":"2017-06-25","proceeding":null,"authors":["NhatHai Phan","Xintao Wu","Dejing Dou"],"abstract":"The remarkable development of deep learning in medicine and healthcare domain\npresents obvious privacy issues, when deep neural networks are built on users'\npersonal and highly sensitive data, e.g., clinical records, user profiles,\nbiomedical images, etc. However, only a few scientific studies on preserving\nprivacy in deep learning have been conducted. In this paper, we focus on\ndeveloping a private convolutional deep belief network (pCDBN), which\nessentially is a convolutional deep belief network (CDBN) under differential\nprivacy. Our main idea of enforcing epsilon-differential privacy is to leverage\nthe functional mechanism to perturb the energy-based objective functions of\ntraditional CDBNs, rather than their results. One key contribution of this work\nis that we propose the use of Chebyshev expansion to derive the approximate\npolynomial representation of objective functions. Our theoretical analysis\nshows that we can further derive the sensitivity and error bounds of the\napproximate polynomial representation. As a result, preserving differential\nprivacy in CDBNs is feasible. We applied our model in a health social network,\ni.e., YesiWell data, and in a handwriting digit dataset, i.e., MNIST data, for\nhuman behavior prediction, human behavior classification, and handwriting digit\nrecognition tasks. Theoretical analysis and rigorous experimental evaluations\nshow that the pCDBN is highly effective. It significantly outperforms existing\nsolutions.","url_abs":"http://arxiv.org/abs/1706.08839v2","url_pdf":"http://arxiv.org/pdf/1706.08839v2.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":"preserving-differential-privacy-in","repo_url":"https://github.com/haiphanNJIT/PrivateDeepLearning","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}},{"paper_slug":"preserving-differential-privacy-in","repo_url":"https://github.com/chuxuantinh/PrivateDeepLearning-master-ct","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}}],"tasks":[],"methods":[{"method_slug":"deep-belief-network","method_name":"Deep Belief Network"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}