Perturbation Towards Easy Samples Improves Targeted Adversarial Transferability
This work addresses the problem of more practical but difficult targeted adversarial attacks in black-box settings for security and robustness applications, representing a strong specific gain rather than a foundational advancement.
The paper tackles the challenge of improving targeted adversarial transferability between neural networks by proposing perturbations towards easy samples in the target class, which avoids high-dimensional density estimation and enhances attack success rates. The result is a method called ESMA that achieves higher success rates than state-of-the-art generative methods while using only 5% of the storage space and less computation time.
The transferability of adversarial perturbations provides an effective shortcut for black-box attacks. Targeted perturbations have greater practicality but are more difficult to transfer between models. In this paper, we experimentally and theoretically demonstrated that neural networks trained on the same dataset have more consistent performance in High-Sample-Density-Regions (HSDR) of each class instead of low sample density regions. Therefore, in the target setting, adding perturbations towards HSDR of the target class is more effective in improving transferability. However, density estimation is challenging in high-dimensional scenarios. Further theoretical and experimental verification demonstrates that easy samples with low loss are more likely to be located in HSDR. Perturbations towards such easy samples in the target class can avoid density estimation for HSDR location. Based on the above facts, we verified that adding perturbations to easy samples in the target class improves targeted adversarial transferability of existing attack methods. A generative targeted attack strategy named Easy Sample Matching Attack (ESMA) is proposed, which has a higher success rate for targeted attacks and outperforms the SOTA generative method. Moreover, ESMA requires only 5% of the storage space and much less computation time comparing to the current SOTA, as ESMA attacks all classes with only one model instead of seperate models for each class. Our code is available at https://github.com/gjq100/ESMA.