AI Paper Atlas

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

Adversarial Ranking Attack and Defense

Mo Zhou, Zhenxing Niu, Le Wang, Qilin Zhang, Gang Hua

ECCV 2020 · Adversarial robustness and security › Black-box and transfer attacks · approach: Robustness benchmarks and defenses · 34 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 maximum-shift-distance adversarial training defense (ablation)ranking model · MNIST, CA+ eps=0.3 w=1mean rank after attack (%): 2.1 -> 30.7Improved
add maximum-shift-distance adversarial training defense (ablation)ranking model · MNIST clean retrievalRecall@1 (%): 99.1 -> 98.3Hurt
add defense (ablation)ranking model · Fashion-MNIST, CA+ eps=0.3 w=1mean rank after attack (%): 1.3 -> 32.5Improved
add defense (ablation)ranking model · Fashion-MNIST clean retrievalRecall@1 (%): 88.8 -> 79.6Hurt
add defense (ablation)ranking model · SOP, CA- eps=0.06 w=1mean rank after attack (%): 100.0 -> 97.3Mixed
add defense (ablation)ranking model · SOP clean retrievalRecall@1 (%): 63.1 -> 46.4Hurt
add defense (ablation)ranking model · SOP, QA- eps=0.06 m=1mean rank after attack (%): 73.8 -> 51.4Improved
raise SP balancing parameter xi from 0 to 100 for QA+ (ablation)SP-QA+ · MNIST m=1mean rank of chosen candidate (%): 0.2 -> 14.1Hurt
raise xi to 10^4 (ablation)SP-QA+ · MNIST m=1mean rank of chosen candidate (%): 14.1 -> 37.9Hurt
use ResNet-18 instead of C2F1 as the ranking modelranking model · MNIST CA- transfer/white-boxmean rank after attack (%): 93.9 -> 66.7Improved
evaluate universal perturbation on unseen imagesI-CA+ · MNISTmean rank (%): 18.1 -> 18.5No clear effect
transfer adversarial candidates from C2F1 to Res18CA+ · MNIST black-boxmean rank (%): 50 -> 31.3Improved

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