AI Paper Atlas

Research map / Adversarial robustness and security / Attacks on text and graphs

Fooling Network Interpretation in Image Classification

Akshayvarun Subramanya, Vipin Pillai, Hamed Pirsiavash

ICCV 2019 · Adversarial robustness and security › Attacks on text and graphs · approach: Adversarial perturbations beyond images · 68 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
add Grad-CAM suppression term to targeted patch loss (ablation)Adversarial patch (VGG19-BN) · ImageNet valenergy ratio: 76.26 -> 6.80Improved
add Grad-CAM suppression term to targeted patch loss (ablation)Adversarial patch (VGG19-BN) · ImageNet valtarget accuracy: 99.98 -> 77.88Hurt
add Grad-CAM suppression term to non-targeted patch loss (ablation)Adversarial patch (VGG19-BN) · ImageNet valenergy ratio: 50.87 -> 2.61Improved
add suppression term, ResNet-34 patch at top-right (ablation)Adversarial patch · ImageNet val 5,000 imagesenergy ratio: 61.9 -> 8.2Improved
add suppression term, DenseNet-121 patch at top-right (ablation)Adversarial patch · ImageNet val 5,000 imagesenergy ratio: 71.3 -> 5.3Improved
encourage Grad-CAM at the opposite corner (uniform heatmap) (ablation)Adversarial patch · ImageNet valtop-left energy: 76.96 -> 14.99Improved
add suppression term on GAIN_ext model (ablation)Adversarial patch · PASCAL VOC-2012energy ratio: 37.90 -> 3.2Improved
evaluate Grad-CAM-trained patch with Occluding Patch (ablation)Adversarial patch · PASCAL VOC-2012 GAIN_extenergy: 80.44 -> 31.59Improved
add Grad-CAM-fooling term to regular PGD examples (ablation)PGD adversarial examples · ImageNet vallocalization error: 67.74 -> 76.02Improved
add Grad-CAM-fooling term to regular PGD examples (ablation)PGD adversarial examples · ImageNet valhistogram intersection: 0.77 -> 0.64Improved
add suppression term, universal patch for class cow (ablation)Universal patch · PASCAL VOC-2012 GAIN_exttarget accuracy: 100 -> 34.3Hurt

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