{"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/better-diffusion-models-further-improve","title":"Better Diffusion Models Further Improve Adversarial Training","arxiv_id":"2302.04638","date":"2023-02-09","proceeding":null,"authors":["Zekai Wang","Tianyu Pang","Chao Du","Min Lin","Weiwei Liu","Shuicheng Yan"],"abstract":"It has been recognized that the data generated by the denoising diffusion probabilistic model (DDPM) improves adversarial training. After two years of rapid development in diffusion models, a question naturally arises: can better diffusion models further improve adversarial training? This paper gives an affirmative answer by employing the most recent diffusion model which has higher efficiency ($\\sim 20$ sampling steps) and image quality (lower FID score) compared with DDPM. Our adversarially trained models achieve state-of-the-art performance on RobustBench using only generated data (no external datasets). Under the $\\ell_\\infty$-norm threat model with $\\epsilon=8/255$, our models achieve $70.69\\%$ and $42.67\\%$ robust accuracy on CIFAR-10 and CIFAR-100, respectively, i.e. improving upon previous state-of-the-art models by $+4.58\\%$ and $+8.03\\%$. Under the $\\ell_2$-norm threat model with $\\epsilon=128/255$, our models achieve $84.86\\%$ on CIFAR-10 ($+4.44\\%$). These results also beat previous works that use external data. We also provide compelling results on the SVHN and TinyImageNet datasets. Our code is available at https://github.com/wzekai99/DM-Improves-AT.","url_abs":"https://arxiv.org/abs/2302.04638v2","url_pdf":"https://arxiv.org/pdf/2302.04638v2.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":"better-diffusion-models-further-improve","repo_url":"https://github.com/wzekai99/dm-improves-at","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"better-diffusion-models-further-improve","repo_url":"https://github.com/BjoernNieth/LS-Dataset-pruning-in-AT","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"better-diffusion-models-further-improve","repo_url":"https://github.com/bbartoldson/adversarial-robustness-limits","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"better-diffusion-models-further-improve","repo_url":"https://github.com/poloclub/robust-principles","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"denoising","task_name":"Denoising"}],"methods":[{"method_slug":"diffusion","method_name":"Diffusion"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":"https://app.syntology.ai/?focus=2302.04638","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}