{"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/differentially-private-releasing-via-deep","title":"Differentially Private Releasing via Deep Generative Model (Technical Report)","arxiv_id":"1801.01594","date":"2018-01-05","proceeding":null,"authors":["Xinyang Zhang","Shouling Ji","Ting Wang"],"abstract":"Privacy-preserving releasing of complex data (e.g., image, text, audio)\nrepresents a long-standing challenge for the data mining research community.\nDue to rich semantics of the data and lack of a priori knowledge about the\nanalysis task, excessive sanitization is often necessary to ensure privacy,\nleading to significant loss of the data utility. In this paper, we present\ndp-GAN, a general private releasing framework for semantic-rich data. Instead\nof sanitizing and then releasing the data, the data curator publishes a deep\ngenerative model which is trained using the original data in a differentially\nprivate manner; with the generative model, the analyst is able to produce an\nunlimited amount of synthetic data for arbitrary analysis tasks. In contrast of\nalternative solutions, dp-GAN highlights a set of key features: (i) it provides\ntheoretical privacy guarantee via enforcing the differential privacy principle;\n(ii) it retains desirable utility in the released model, enabling a variety of\notherwise impossible analyses; and (iii) most importantly, it achieves\npractical training scalability and stability by employing multi-fold\noptimization strategies. Through extensive empirical evaluation on benchmark\ndatasets and analyses, we validate the efficacy of dp-GAN.","url_abs":"http://arxiv.org/abs/1801.01594v2","url_pdf":"http://arxiv.org/pdf/1801.01594v2.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":"differentially-private-releasing-via-deep","repo_url":"https://github.com/alps-lab/dpgan","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":{"status":"unanswered"}},{"paper_slug":"differentially-private-releasing-via-deep","repo_url":"https://github.com/alexandrehuat/dp-gan","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"privacy-preserving","task_name":"Privacy Preserving"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1801.01594","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}