AI Paper Atlas

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

Adversarial Attack Type I: Cheat Classifiers by Significant Changes

Sanli Tang, Xiaolin Huang, Mingjian Chen, Chengjin Sun, Jie Yang

TPAMI 2019 · Adversarial robustness and security › Black-box and transfer attacks · approach: Transfer attacks on face recognition · 37 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
apply feature squeezing detection to Type II (FGSM) instead of Type IMLP feature squeezing, zeta 0.2510 · MNISTdetection rate: 0.870 vs 0.306Hurt
use Type I SVAE attack instead of clean samplesMLP feature squeezing, zeta 0.6089 · MNISTdetection rate: 0.102 -> 0.122No clear effect
use Type I SVAE attack instead of clean samplesFaceNet feature squeezing, 5x5 · CelebAdetection rate: 0.064 -> 0.032No clear effect
use Type II PGD instead of Type I SVAE examplesFaceNet feature squeezing, 5x5 · CelebAdetection rate: 0.032 -> 0.971Improved
adversarial logit pairing training (ablation)LeNet, Type II attack · MNISTaccuracy: 98.5%Improved
adversarial logit pairing training (ablation)LeNet, Type I attack · MNISTaccuracy: 14.2%Hurt
strengthen FaceNet by FGSM adversarial training (ablation)FaceNet, PGD attack · CelebA/LFWaccuracy: 0.996Improved
strengthen FaceNet by FGSM adversarial training (ablation)FaceNet, Type I attack · CelebA/LFWaccuracy: 0.110No clear effect
strengthen FaceNet by PGD adversarial training (ablation)FaceNet, Type I attack · CelebA/LFWaccuracy: 0.140No clear effect
strengthen FaceNet by SVAE Type I adversarial training (ablation)FaceNet, Type I attack · CelebA/LFWaccuracy: 0.554 vs 0.110Improved

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