AI Paper Atlas

Research map / Label-efficient and robust learning / Self-supervised and contrastive

Detecting Backdoors in Pre-Trained Encoders

Shiwei Feng, Guanhong Tao, Siyuan Cheng, Guangyu Shen, Xiangzhe Xu, Yingqi Liu, Kaiyuan Zhang, Shiqing Ma, Xiangyu Zhang

CVPR 2023 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Self-supervision in supervised pipelines · 48 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 minimal-trigger inversion to detect backdoored encodersDECREE · Encoders across GTSRB, SVHN, and STL-10 attack targetsdetection accuracy: 99.1; 97.7; 99.1Improved
Use a shadow dataset instead of the pre-training dataset for trigger inversionDECREE · CIFAR10-pretrained encoders, STL-10 shadow datasetseparation of clean and backdoored encoders by PL1-NormImproved
Add a similarity-loss term that encourages triggered inputs to be less similarOriginal SSL backdoor attack · CIFAR10 encoder, SVHN target, downstream STL-10attack success rate: 99.9% to 69.9%Hurt
Add a similarity-loss term that encourages triggered inputs to be less similarOriginal SSL backdoor attack · CIFAR10 encoder, SVHN targetPL1-Norm: 0.14Mixed
Use Text-on-Pair attacks with natural-language target captionsDECREE · CLIP, SVHN attack targetdetection accuracy: 87.5Hurt

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