AI Paper Atlas

Research map / Adversarial robustness and security / Backdoor and poisoning

One-Pixel Signature: Characterizing CNN Models for Backdoor Detection

Shanjiaoyang Huang, Weiqi Peng, Zhiwei Jia, Zhuowen Tu

ECCV 2020 · Adversarial robustness and security › Backdoor and poisoning · approach: Training-time backdoor defense · 63 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 one-pixel signature detector instead of Neural Cleansebackdoor detector · MNISTdetection rate: 60.53% -> 95.72%Improved
use one-pixel signature detector instead of Neural Cleansebackdoor detector · fashion-MNISTdetection rate: 63.16% -> 94.59%Improved
use one-pixel signature detector instead of Neural Cleansebackdoor detector · GTSRBdetection rate: 60.71% -> 88.47%Improved
use one-pixel signature detector instead of Neural Cleansebackdoor detector · ImageNet (10-class subset)detection rate: 54.42% -> 88.69%Improved
use one-pixel signature detector instead of ABSbackdoor detector · GTSRBdetection rate: 66.33% -> 88.47%Improved
test on unseen architecture without training on itOne-Pixel Signature detector · MNISTdetection rate: 95.72% -> 76.50 ± 12.07 (lowest, ResNet unseen)Hurt
train on 175+175 models instead of 300 pairs (ablation)One-Pixel Signature detector · GTSRBdetection rate: 88.47% -> 87.31%No clear effect
add per-channel samples as training examples (ablation)One-Pixel Signature detector · GTSRBtraining pairs needed: 300 pairs -> 15 pairs for same rateImproved
add per-channel samples as training examples (ablation)One-Pixel Signature detector · GTSRBdetection rate: 97% with 100 pairsImproved
use one-pixel signature instead of Neural Cleanse/ABS on all-to-all attackbackdoor detector · all-to-all attackdetection success rate: around 50% -> above 95%Improved
use all-to-one attack rather than one-to-oneOne-Pixel Signature detector · all-to-one attackdetection rateNo clear effect
default image constant 0 versus average testing image (ablation)One-Pixel Signature · MNIST (LeNet-5)detection rateNo clear effect

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