ML Security PapersKeep It Real: Challenges in Attacking Compression-Based Adversarial Purification
Samuel Räber, Till Aczel, Andreas Plesner et al. · ETH Zürich
Rigorous adaptive attack evaluation of compression-based purification shows high-realism reconstructions resist attacks via distributional alignment, not gradient masking
Previous work has suggested that preprocessing images through lossy compression can defend against adversarial perturbations, but comprehensive attack evaluations have been lacking. In this paper, we construct strong white-box and adaptive attacks against various compression models and identify a critical challenge for attackers: high realism in reconstructed images significantly increases attack difficulty. Through rigorous evaluation across multiple attack scenarios, we demonstrate that compression models capable of producing realistic, high-fidelity reconstructions are substantially more resistant to our attacks. In contrast, low-realism compression models can be broken. Our analysis reveals that this is not due to gradient masking. Rather, realistic reconstructions maintaining distributional alignment with natural images seem to offer inherent robustness. This work highlights a significant obstacle for future adversarial attacks and suggests that developing more effective techniques to overcome realism represents an essential challenge for comprehensive security evaluation.