Better Diffusion Models Further Improve Adversarial Training. (arXiv:2302.04638v1 [cs.CV])

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 …

adversarial data development diffusion models efficiency generated higher quality question rapid score sim training