Research map / Label-efficient and robust learning / OOD and anomaly detection
Bypassing the Transport Plan: Dynamic Reweighting for Out-of-Distribution Detection with Optimal Transport
Yang Xiao, Weiming Liu, Jun Dan, Tengyue Xu, Fan Wang, Hua Yu, Junhao Dong, Jiao Liu, Shunjie Dong, Lianyong Qi
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 changed | Where | Measured | Effect |
|---|---|---|---|
| replace max-softmax OOD scoring with DREW-supervised OOD head (ablation) | Fixmatch · CIFAR-100 (55 known, 50 labels/class) | AUROC: 57.3 -> 91.7 | Improved |
| replace POT redundant-mass OT with dynamic reweighting (ablation) | FixMatch+POT · CIFAR-100 (55 known, 50 labels/class) | AUROC: 88.4 -> 91.7 | Improved |
| replace POT with DREW on CIFAR-10 AUROC | POT · CIFAR-10 (6 known classes) | AUROC: 99.7 -> 99.6 (50 labels/class) | No clear effect |
| use exact EMD solver instead of Sinkhorn in DREW (ablation) | DREW · CIFAR-100 (55 known, 50 labels/class) | AUROC: 91.2 -> 91.5 | Improved |
| plug DREW into any OT solver vs POT (ablation) | POT · CIFAR-100 (55 known, 50 labels/class) | AUROC: 90.1 -> 91.0 | Improved |
| remove need for redundant mass k (DREW vs POT with k=1.25) (ablation) | POT · CIFAR-100 (55 known, 50 labels/class) | pseudo-label AUROC: 54.9 -> 59.7 | Improved |
| vary tau_a from 0.001 to 1 (ablation) | DREW · CIFAR-100 | accuracy/AUROC | No clear effect |
| set OOD loss weight gamma too small (0.01) (ablation) | DREW · CIFAR-100 | OOD detection quality | Hurt |
| DREW vs POT on ImageNet-30 | POT · ImageNet-30 (20 known classes) | AUROC: 97.4 -> 97.6 | No clear effect |
| approximate SemiUOT with smooth log-sum-exp for training speed | DREW · Training efficiency | s/iteration | Improved |
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.
- OpenMatch — OpenMatch: Open-Set Semi-supervised Learning with Open-set Consistency Regularization (NeurIPS 2021) · better on 2 of 2 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
- Can We Ignore Labels in Out of Distribution Detection? (ICLR 2025 · similarity 0.89)
- Image Background Serves as Good Proxy for Out-of-distribution Data (ICLR 2024 · similarity 0.88)
- Expecting The Unexpected: Towards Broad Out-Of-Distribution Detection (NeurIPS 2024 · similarity 0.87)
- Enhancing the Power of OOD Detection via Sample-Aware Model Selection (CVPR 2024 · similarity 0.87)
- The Best of Both Worlds: On the Dilemma of Out-of-distribution Detection (NeurIPS 2024 · similarity 0.87)
- Double Descent Meets Out-of-Distribution Detection: Theoretical Insights and Empirical Analysis on the Role of Model Complexity (NeurIPS 2025 · similarity 0.87)
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.
