Eigenpatches -- Adversarial Patches from Principal Components
This work addresses the problem of reducing computational costs for adversarial attacks in computer vision, though it appears incremental as it builds on existing patch-based methods.
The paper tackled the computational expense of generating adversarial patches for object detectors by analyzing 375 patches and using their principal components to create effective 'eigenpatches' that successfully fool detections.
Adversarial patches are still a simple yet powerful white box attack that can be used to fool object detectors by suppressing possible detections. The patches of these so-called evasion attacks are computational expensive to produce and require full access to the attacked detector. This paper addresses the problem of computational expensiveness by analyzing 375 generated patches, calculating the principal components of these and show, that linear combinations of the resulting "eigenpatches" can be used to fool object detections successfully.