AI Paper Atlas

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

Diffusion Models for Imperceptible and Transferable Adversarial Attack

Jianqi Chen, Hao Chen, Keyan Chen, Yilan Zhang, Zhengxia Zou, Zhenwei Shi

TPAMI 2024 · Adversarial robustness and security › Black-box and transfer attacks · approach: Transferable adversarial examples · 160 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
perturb diffusion latent instead of text embedding (ablation)DiffAttack · ImageNet-Compatible, surrogate Inc-v3AVG(w/o self) accuracy: 80.1 -> 65.4Improved
perturb latent instead of text (ablation)DiffAttack · ImageNet-Compatible, surrogate Inc-v3FID: 58.8 -> 62.3Hurt
add hard pseudo mask restricting perturbation to foreground (nearest) (ablation)DiffAttack · ImageNet-Compatible, surrogate Inc-v3AVG(w/o self) accuracy: 65.4 -> 74.2Hurt
add hard pseudo mask restricting perturbation to foreground (nearest) (ablation)DiffAttack · ImageNet-Compatible, surrogate Inc-v3FID: 62.3 -> 59.1Improved
add soft mask with bilinear upsampling (ablation)DiffAttack · ImageNet-Compatible, surrogate Inc-v3AVG(w/o self) accuracy: 65.4 -> 76.5Hurt
use top-2 category names as guided text instead of top-1 (ablation)DiffAttack · ImageNet-Compatible, surrogate Inc-v3AVG(w/o self) accuracy: 65.4 -> 70.4Hurt
use top-5 category names as guided text instead of top-1 (ablation)DiffAttack · ImageNet-Compatible, surrogate Inc-v3AVG(w/o self) accuracy: 65.4 -> 69.1Hurt
add structure controls (inversion strength and self-attention) (ablation)DiffAttack · ImageNet-Compatible, surrogate VGG-19FID: 96.2 -> 63.9Improved
add structure controls (inversion strength and self-attention) (ablation)DiffAttack · ImageNet-Compatible, surrogate VGG-19AVG(w/o self) accuracy: 25.5 -> 38.2Hurt
use DiffAttack versus S2I-FGSM on VGG-19 surrogatetransfer attack · ImageNet-Compatible, surrogate VGG-19AVG(w/o self) accuracy: 49.0 -> 38.2Improved
use DiffAttack versus MI-FGSM on Mob-v2 surrogatetransfer attack · ImageNet-Compatible, surrogate Mob-v2AVG(w/o self) accuracy: 49.9 -> 40.5Improved
use DiffAttack versus MI-FGSM on Inc-v3 surrogatetransfer attack · ImageNet-Compatible, surrogate Inc-v3AVG(w/o self) accuracy: 60.6 -> 65.4Hurt

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