Boosting Cross-task Transferability of Adversarial Patches with Visual Relations
This addresses the robustness evaluation of foundational multi-task AI systems like Visual ChatGPT, though it is incremental as it builds on existing adversarial patch methods.
The paper tackles the problem of limited cross-task transferability of adversarial examples in multi-task AI systems, proposing VRAP, a method that uses visual relations to generate adversarial patches, which significantly outperforms previous methods in black-box transferability across visual reasoning tasks.
The transferability of adversarial examples is a crucial aspect of evaluating the robustness of deep learning systems, particularly in black-box scenarios. Although several methods have been proposed to enhance cross-model transferability, little attention has been paid to the transferability of adversarial examples across different tasks. This issue has become increasingly relevant with the emergence of foundational multi-task AI systems such as Visual ChatGPT, rendering the utility of adversarial samples generated by a single task relatively limited. Furthermore, these systems often entail inferential functions beyond mere recognition-like tasks. To address this gap, we propose a novel Visual Relation-based cross-task Adversarial Patch generation method called VRAP, which aims to evaluate the robustness of various visual tasks, especially those involving visual reasoning, such as Visual Question Answering and Image Captioning. VRAP employs scene graphs to combine object recognition-based deception with predicate-based relations elimination, thereby disrupting the visual reasoning information shared among inferential tasks. Our extensive experiments demonstrate that VRAP significantly surpasses previous methods in terms of black-box transferability across diverse visual reasoning tasks.