On the Robustness of Quality Measures for GANs

Autor:	Alfarra, Motasem, Pérez, Juan C., Frühstück, Anna, Torr, Philip H. S., Wonka, Peter, Ghanem, Bernard
Rok vydání:	2022
Předmět:	Computer Science - Machine Learning
Druh dokumentu:	Working Paper
Popis:	This work evaluates the robustness of quality measures of generative models such as Inception Score (IS) and Fr\'echet Inception Distance (FID). Analogous to the vulnerability of deep models against a variety of adversarial attacks, we show that such metrics can also be manipulated by additive pixel perturbations. Our experiments indicate that one can generate a distribution of images with very high scores but low perceptual quality. Conversely, one can optimize for small imperceptible perturbations that, when added to real world images, deteriorate their scores. We further extend our evaluation to generative models themselves, including the state of the art network StyleGANv2. We show the vulnerability of both the generative model and the FID against additive perturbations in the latent space. Finally, we show that the FID can be robustified by simply replacing the standard Inception with a robust Inception. We validate the effectiveness of the robustified metric through extensive experiments, showing it is more robust against manipulation. Comment: Accepted at the European Conference in Computer Vision (ECCV 2022)
Databáze:	arXiv
Externí odkaz:	http://arxiv.org/abs/2201.13019 Zobrazit plný text záznamu View this record from Arxiv