AI Paper Atlas

Research map / Adversarial robustness and security / Backdoor and poisoning

Towards Stable Backdoor Purification through Feature Shift Tuning

Rui Min, Zeyu Qin, Li Shen, Minhao Cheng

NeurIPS 2023 · Adversarial robustness and security › Backdoor and poisoning · approach: Training-time backdoor defense · 5 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 Feature Shift Tuning instead of the strongest compared defenseBackdoor purification on CIFAR-10, GTSRB, and Tiny-ImageNet · CIFAR-10, GTSRB, Tiny-ImageNetASR: 34.04% and 26.91%Improved
Use Feature Shift Tuning instead of ANP or I-BAUBackdoor purification on CIFAR-10 and GTSRB · CIFAR-10 and GTSRBASR: 11.34% and 32.16%Improved
Add feature-extractor tuning with a reinitialized, frozen classifier (FE-tuning) rather than LP (ablation)Backdoor purification · CIFAR-10, BadNet, Blended, SSBA, and LC at 1% and 0.5% poisoningASR: 90.91 -> 15.11Improved
Add whole-model tuning with a reinitialized classifier (FT-init) rather than vanilla FT (ablation)Backdoor purification · CIFAR-10, BadNet, Blended, SSBA, and LC at 1% and 0.5% poisoningASR: 74.36 -> 52.60Improved
Use FST rather than FE-tuning (ablation)Backdoor purification · Tiny-ImageNetC-Acc: 8.1%Improved
Add the classifier weight-deviation penalty and projection constraint (FST) rather than FT-init (ablation)Backdoor purification · CIFAR-10 and GTSRBASRImproved
Add the classifier weight-deviation penalty and projection constraint (FST) rather than FT-init (ablation)Backdoor purification · CIFAR-100 and Tiny-ImageNetASR: 32.31%Mixed
Add the projection constraint (ablation)Feature Shift Tuning · CIFAR-10, BadNet and Blendedconvergence stabilityImproved
Increase α from 0 to 0.1, then continue increasing it (ablation)Feature Shift Tuning · CIFAR-10, BadNet and BlendedASR: ASR < 5%; ASR < 4%; accuracy degradation < 1.5%Improved
Reduce the tuning set to 0.1% of training data (ablation)Feature Shift Tuning · CIFAR-10ASR: less than 2%No clear effect
Tune for only a few epochs with FST rather than other tuning defenses (ablation)Backdoor purification · CIFAR-10, 1% poisoningASR: below 5%Improved
Extend FST's weight-deviation penalty from the last classifier layer to all linear classifier layers (ablation)FST on VGG19-BN · CIFAR-10ASRImproved

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