{"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-generative-adversarial-1","title":"Differentially Private Generative Adversarial Network","arxiv_id":"1802.06739","date":"2018-02-19","proceeding":null,"authors":["Liyang Xie","Kaixiang Lin","Shu Wang","Fei Wang","Jiayu Zhou"],"abstract":"Generative Adversarial Network (GAN) and its variants have recently attracted\nintensive research interests due to their elegant theoretical foundation and\nexcellent empirical performance as generative models. These tools provide a\npromising direction in the studies where data availability is limited. One\ncommon issue in GANs is that the density of the learned generative distribution\ncould concentrate on the training data points, meaning that they can easily\nremember training samples due to the high model complexity of deep networks.\nThis becomes a major concern when GANs are applied to private or sensitive data\nsuch as patient medical records, and the concentration of distribution may\ndivulge critical patient information. To address this issue, in this paper we\npropose a differentially private GAN (DPGAN) model, in which we achieve\ndifferential privacy in GANs by adding carefully designed noise to gradients\nduring the learning procedure. We provide rigorous proof for the privacy\nguarantee, as well as comprehensive empirical evidence to support our analysis,\nwhere we demonstrate that our method can generate high quality data points at a\nreasonable privacy level.","url_abs":"http://arxiv.org/abs/1802.06739v1","url_pdf":"http://arxiv.org/pdf/1802.06739v1.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-generative-adversarial-1","repo_url":"https://github.com/illidanlab/dpgan","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}},{"paper_slug":"differentially-private-generative-adversarial-1","repo_url":"https://github.com/Pushkar-v/Generating-Synthetic-Data-using-GANs","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}}],"tasks":[{"task_slug":null,"task_name":"Generative Adversarial Network"}],"methods":[{"method_slug":"convolution","method_name":"Convolution"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1802.06739","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}