AI Paper Atlas

Research map / Adversarial robustness and security / Black-box and transfer attacks

PatchAttack: A Black-box Texture-based Attack with Reinforcement Learning

Chenglin Yang, Adam Kortylewski, Cihang Xie, Yinzhi Cao, Alan Yuille

ECCV 2020 · Adversarial robustness and security › Black-box and transfer attacks · approach: Robustness benchmarks and defenses · 109 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
optimize patch color (RGB) instead of gray in monochrome patch attack (ablation)MPA · ImageNet non-targeted, averaged over 4 networksavg occluded area (%): 6.67 -> 5.37Improved
add texture from class-specific dictionary to patches (TPA vs MPA_RGB) (ablation)MPA_RGB · ImageNet targeted, ResNet50target accuracy (%): 25.90 -> 99.70Improved
use reinforcement learning instead of Hastings sampling in the same RGB space (ablation)HPA_RGB · ImageNet non-targeted, ResNet50avg occluded area (%): 16.88 -> 5.41Improved
use reinforcement learning instead of Hastings sampling in texture space (ablation)HPA_N10_4% · ImageNet targeted, ResNet50avg queries: 14345 -> 8643Improved
use reinforcement learning instead of Hastings sampling in texture space (ablation)HPA_N10_4% · ImageNet targeted, ResNet50target accuracy (%): 99.80 -> 99.70No clear effect
increase patch budget from 2% to 4% of image area, 10 patches (ablation)TPA · ImageNet targeted, ResNet50target accuracy (%): 97.60 -> 99.70Improved
increase patch budget from 4% to 10% of image area, 10 patches (ablation)TPA · ImageNet targeted, ResNet50avg queries: 8643 -> 3747Improved
use up-to-8 patches of 2% instead of up-to-4 patches of 4% (ablation)TPA · ImageNet non-targeted, ResNet50avg occluded area (%): 5.06 -> 3.10Improved
add RL monochrome patch attack to Denoise-ResNet152 defense, compare to normal ResNet50MPA_RGB · ImageNet non-targetedavg occluded area (%): 5.41 -> 0.48Improved
shape-biased network as a defense against TPATPA_N10_4% · ImageNet targetedtarget accuracy (%): 99.70 -> 100.00No clear effect
apply Local Gradients Smoothing defenseTPA_N10_4% · ImageNet targeted, ResNet50target accuracy (%): 99.70 -> 97.50No clear effect
use white-box adversarial patch dictionary from VGG19 on other networksAdvPatchDict · ImageNet targetedtarget accuracy (%): 98.20 -> 0.00Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy