AI Paper Atlas

Research map / Label-efficient and robust learning / OOD and anomaly detection

Synthesizing Near-Boundary OOD Samples for Out-of-Distribution Detection

Jinglun Li, Kaixun Jiang, Zhaoyu Chen, Bo Lin, Yao Tang, Weifeng Ge, Wenqiang Zhang

ICCV 2025 · Label-efficient and robust learning › OOD and anomaly detection · approach: Spurious-aware OOD detection · 4 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 iterative gradient-guided generation instead of direct negative-label prompts (ablation)SynOOD · ImageNet-1k avg 4 OOD setsAUROC: 94.85 -> 96.21Improved
use iterative gradient-guided generation instead of direct negative-label prompts (ablation)SynOOD · ImageNet-1k avg 4 OOD setsFPR95: 22.59 -> 17.29Improved
add negative label feature fine-tuning to gradient-generated data (ablation)SynOOD · ImageNet-1k avg 4 OOD setsAUROC: 96.21 -> 97.01Improved
add negative label feature fine-tuning to gradient-generated data (ablation)SynOOD · ImageNet-1k avg 4 OOD setsFPR95: 17.29 -> 14.27Improved
add negative label fine-tuning when using direct negative-label images (ablation)SynOOD · ImageNet-1k avg 4 OOD setsAUROC: 94.85 -> 95.01Improved
add projection layer trained on direct negative-label images (ablation)NegLabel · ImageNet-1k avg 4 OOD setsAUROC: 94.21 -> 94.85Improved
add projection layer trained on gradient-generated images (ablation)NegLabel · ImageNet-1k avg 4 OOD setsFPR95: 25.40 -> 17.29Improved
train projection and label features separately instead of jointly (ablation)SynOOD · ImageNet-1k avg 4 OOD setsFPR95: 19.41 -> 14.27Improved
exclude the trained projection layer during label feature fine-tuning (ablation)SynOOD · ImageNet-1k avg 4 OOD setsAUROC: 96.49 -> 97.01Improved
increase synthetic sample count m beyond 50k (ablation)SynOOD · ImageNet-1k avg 4 OOD setsAUROCHurt
apply SynOOD on CLIP ResNet50 versus NegLabelSynOOD · ImageNet-1k avg 4 OOD setsAUROC: 92.97 -> 95.80Improved
apply SynOOD on CLIP ViT-B/32 versus NegLabelSynOOD · ImageNet-1k avg 4 OOD setsAUROC: 93.67 -> 95.93Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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