Research map / Label-efficient and robust learning / OOD and anomaly detection
Mixture Prototype Flow Matching for Open-Set Supervised Anomaly Detection
Fuyun Wang, Yuanzhi Wang, Xu Guo, Sujia Huang, Tong Zhang, Dan Wang, Hui Yan, Xin Liu, Zhen Cui
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 |
|---|---|---|---|
| add Gaussian mixture prototype flow learning (MPFL) (ablation) | MPFM · nine benchmarks, general and hard | AUC | Improved |
| add mutual information maximization regularizer (MIMR) (ablation) | MPFM · nine benchmarks, general and hard | AUC | Improved |
| use both MPFL and MIMR together (ablation) | MPFM · nine benchmarks | AUC | Improved |
| add global likelihood scorer Mg (ablation) | MPFM · general setting, 10 anomalies | AUC | Improved |
| add residual anomaly scorer Mr (ablation) | MPFM · general setting, 10 anomalies | AUC | Improved |
| increase GMM components from 8 to 32 (ablation) | MPFM · AITEX, general, 10 anomalies | AUC: 0.980 -> 0.992 | Improved |
| increase GMM components from 32 to 64 (ablation) | MPFM · AITEX, general, 10 anomalies | AUC: 0.992 -> 0.989 | Hurt |
| set MIMR weight to 0.1 rather than 0.001 (ablation) | MPFM · MVTec AD, general, 10 anomalies | AUC: 0.972 -> 0.982 | Improved |
| increase MIMR weight from 0.1 to 1.0 (ablation) | MPFM · AITEX, general, 10 anomalies | AUC: 0.992 -> 0.981 | Hurt |
| replace unimodal Gaussian prototypes (DPDL) with mixture prototype flow (MPFM) | DPDL · Mastcam, general, 10 anomalies | AUC: 0.934 -> 0.968 | Improved |
| replace unimodal Gaussian prototypes (DPDL) with mixture prototype flow (MPFM) | DPDL · Mastcam, hard, 1 anomaly | AUC: 0.733 -> 0.760 | Improved |
Closest papers by meaning
- Scalable, Explainable and Provably Robust Anomaly Detection with One-Step Flow Matching (NeurIPS 2025 · similarity 0.88)
- Distribution Prototype Diffusion Learning for Open-set Supervised Anomaly Detection (CVPR 2025 · similarity 0.88)
- Anomaly Heterogeneity Learning for Open-set Supervised Anomaly Detection (CVPR 2024 · similarity 0.86)
- DFM: Differentiable Feature Matching for Anomaly Detection (CVPR 2025 · similarity 0.86)
- Generating and Reweighting Dense Contrastive Patterns for Unsupervised Anomaly Detection (AAAI 2024 · similarity 0.85)
- Stream-level Flow Matching with Gaussian Processes (ICML 2025 · similarity 0.85)
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.
