{"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/improving-generalization-and-stability-of","title":"Improving Generalization and Stability of Generative Adversarial Networks","arxiv_id":"1902.03984","date":"2019-02-11","proceeding":"ICLR 2019 5","authors":["Hoang Thanh-Tung","Truyen Tran","Svetha Venkatesh"],"abstract":"Generative Adversarial Networks (GANs) are one of the most popular tools for\nlearning complex high dimensional distributions. However, generalization\nproperties of GANs have not been well understood. In this paper, we analyze the\ngeneralization of GANs in practical settings. We show that discriminators\ntrained on discrete datasets with the original GAN loss have poor\ngeneralization capability and do not approximate the theoretically optimal\ndiscriminator. We propose a zero-centered gradient penalty for improving the\ngeneralization of the discriminator by pushing it toward the optimal\ndiscriminator. The penalty guarantees the generalization and convergence of\nGANs. Experiments on synthetic and large scale datasets verify our theoretical\nanalysis.","url_abs":"http://arxiv.org/abs/1902.03984v1","url_pdf":"http://arxiv.org/pdf/1902.03984v1.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":"improving-generalization-and-stability-of","repo_url":"https://github.com/htt210/GeneralizationAndStabilityInGANs","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[],"methods":[{"method_slug":"convolution","method_name":"Convolution"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1902.03984","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}