AI Paper Atlas

Research map / Adversarial robustness and security / Backdoor and poisoning

Test-Time Poisoned Sample Detection by Exploiting Shallow Malicious Matching in Backdoored CLIP

Zhengyao Song, Meixi Zheng, Ke Xu, Yongqiang Li, Baoyuan Wu

ICLR 2026 · Adversarial robustness and security › Backdoor and poisoning · approach: Detecting poisoned samples

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
sample along the positive direction instead of the negative direction (ablation)Subspace Detection · ImageNet-1K, BadNetsAUROC: 0.525 -> 0.962Improved
sample along the positive direction instead of the negative direction (ablation)Subspace Detection · ImageNet-1K, Carlini & TerzisAUROC: 0.702 -> 0.994Improved
sample along the positive direction instead of the negative direction (ablation)Subspace Detection · ImageNet-1K, WaNetAUROC: 0.246 -> 0.931Improved
combine all three text transformations instead of the weakest single one (ablation)Subspace Detection · ImageNet-1K, WaNetAUROC: 0.543 -> 0.931Improved
combine all three text transformations instead of the weakest single one (ablation)Subspace Detection · ImageNet-1K, BadNetsAUROC: 0.749 -> 0.962Improved
add Font to Description (ablation)Subspace Detection · ImageNet-1K, WaNetAUROC: 0.543 -> 0.913Improved
increase augmented text features n from 18 to 90 (ablation)Subspace Detection · ImageNet-1K, BadNetsAUROC: 0.925 -> 0.972Improved
increase sampled text features ns from 9 to 21 (ablation)Subspace Detection · ImageNet-1K, BadNetsAUROC: 0.925 -> 0.972Improved
increase mixture components L from 1 to 4 (ablation)Subspace Detection · ImageNet-1K (7 attacks)AUROCImproved
use manifold approximation subspace instead of a single transformation with Euclidean distance (ablation)Subspace Detection · ImageNet-1K, WaNetAUROC: 0.421 -> 0.931Improved
use manifold approximation subspace instead of a single transformation with Euclidean distance (ablation)Subspace Detection · ImageNet-1K, BadCLIPAUROC: 0.593 -> 0.966Improved
use L2 distance instead of L1 distance (ablation)Subspace Detection · ImageNet-1K, BadNetAUROC: 0.671 -> 0.962Improved

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