Image Shortcut Squeezing: Countering Perturbative Availability Poisons with Compression

TL;DR

This paper demonstrates that simple image compression techniques like ISS can effectively counter perturbative availability poisons, challenging the belief that no practical defenses exist against such attacks.

Contribution

The study introduces Image Shortcut Squeezing (ISS), a simple compression-based method that significantly outperforms existing countermeasures against PAPs and offers higher generalizability and efficiency.

Findings

ISS restores CIFAR-10 accuracy to 81.73%

ISS surpasses previous preprocessing defenses by 37.97%

ISS outperforms adversarial training and resists adaptive poisoning

Abstract

Perturbative availability poisons (PAPs) add small changes to images to prevent their use for model training. Current research adopts the belief that practical and effective approaches to countering PAPs do not exist. In this paper, we argue that it is time to abandon this belief. We present extensive experiments showing that 12 state-of-the-art PAP methods are vulnerable to Image Shortcut Squeezing (ISS), which is based on simple compression. For example, on average, ISS restores the CIFAR-10 model accuracy to $81.73\%$, surpassing the previous best preprocessing-based countermeasures by $37.97\%$ absolute. ISS also (slightly) outperforms adversarial training and has higher generalizability to unseen perturbation norms and also higher efficiency. Our investigation reveals that the property of PAP perturbations depends on the type of surrogate model used for poison generation, and it explains why a specific ISS compression yields the best performance for a specific type of PAP perturbation. We further test stronger, adaptive poisoning, and show it falls short of being an ideal defense against ISS. Overall, our results demonstrate the importance of considering various (simple) countermeasures to ensure the meaningfulness of analysis carried out during the development of PAP methods.