AI Paper Atlas

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

Transferable Adversarial Perturbations

Wen Zhou, Xin Hou, Yongjun Chen, Mengyun Tang, Xiangqi Huang, Xiang Gan, Yong Yang

ECCV 2018 · Adversarial robustness and security › Black-box and transfer attacks · approach: Transferable adversarial examples · 267 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
use TAP instead of BIM to craft perturbations on Inception-V3black-box attack on Inception-V4 · ImageNet-compatible, source Inception-V3recognition accuracy: 67.7% -> 22.1%Improved
use TAP instead of FGSM on Inception-V3black-box attack on Inception-ResNet-V2 · ImageNet-compatible, source Inception-V3recognition accuracy: 72.9% -> 24.7%Improved
use TAP instead of BIM on Inception-V3adversarially trained adv-v3 · ImageNet-compatible, source Inception-V3recognition accuracy: 82.4% -> 52.8%Improved
add the intermediate feature-distance term (lambda 0.05, eta 0) (ablation)BIM · Inception-V3 source, standard and ensemble modelsrecognition accuracyImproved
add the feature-distance term (lambda 0.05, eta 0) (ablation)BIM · adversarially trained modelsrecognition accuracyImproved
add the smooth regularization term (lambda 0, eta 10^3) (ablation)BIM · adversarially trained modelsrecognition accuracyImproved
add the smooth regularization term (lambda 0, eta 10^3) (ablation)BIM · Inception-V3 white-boxrecognition accuracyHurt
add the smooth regularization term (lambda 0, eta 10^3) (ablation)BIM · Inception-V4 and Inception-ResNet-V2recognition accuracyHurt
increase lambda (ablation)TAP · adversarially trained vs standard modelsrecognition accuracyMixed
increase eta (ablation)TAP · adversarially trained vs standard modelsrecognition accuracyMixed
set power normalization alpha to 0.5 instead of other values (ablation)TAP · defense modelsrecognition accuracyImproved
increase smoothing kernel size s above 3 (ablation)TAP · all modelsrecognition accuracyHurt

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