Research map / Label-efficient and robust learning / OOD and anomaly detection
Mixture of Gaussian-Distributed Prototypes With Generative Modelling for Interpretable and Trustworthy Image Recognition
Chong Wang, Yuanhong Chen, Fengbei Liu, Y. Liu, Davis James McCarthy, Helen M. L. Frazer, Gustavo Carneiro
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 auxiliary loss L_aux (ablation) | MGProto · CUB (R34) | accuracy: 79.6 -> 81.1 | Improved |
| add prototype mining L_mn (ablation) | MGProto · CUB (R34) | accuracy: 79.6 -> 81.5 | Improved |
| add both L_aux and L_mn (ablation) | MGProto · Cars (R50) | accuracy: 87.0 -> 89.2 | Improved |
| add L_mn (ablation) | MGProto · CUB interpretability V19 | DAUC: 3.91 -> 3.20 | Improved |
| add L_aux (ablation) | MGProto · CUB interpretability V19 | consistency: 92.97 -> 92.90 | No clear effect |
| add prototype mining | ProtoPNet · CUB (ResNet34) | accuracy: 72.4 -> 76.5 | Improved |
| add prototype mining | TesNet · CUB (ResNet34) | accuracy: 76.2 -> 78.6 | Improved |
| add prototype mining | ST-ProtoPNet · CUB (ResNet34) | accuracy: 77.4 -> 79.3 | Improved |
| use diversity constraint instead of closed-form M-step (ablation) | MGProto · CUB (R34) | accuracy: 81.2 -> 82.2 | Improved |
| use diversity constraint instead of closed-form M-step (ablation) | MGProto · Cars (R50) | accuracy: 88.3 -> 89.2 | Improved |
| importance-based pruning of 20% prototypes | MGProto · CUB (ResNet34) | accuracy: 82.2 -> 82.4 | Improved |
| importance-based pruning of 90% prototypes | MGProto · CUB (ResNet34) | accuracy: 82.2 -> 80.5 | Hurt |
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.
- PIP-Net — PIP-Net: Patch-Based Intuitive Prototypes for Interpretable Image Classification (CVPR 2023) · better on 3 of 3 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
- Interpretable Image Classification via Non-parametric Part Prototype Learning (CVPR 2025 · similarity 0.90)
- ProtoArgNet: Interpretable Image Classification with Super-Prototypes and Argumentation (AAAI 2025 · similarity 0.87)
- Improving Prototypical Visual Explanations with Reward Reweighing, Reselection, and Retraining (ICML 2024 · similarity 0.87)
- Post-hoc Part-Prototype Networks (ICML 2024 · similarity 0.86)
- Interpretable Image Classification with Adaptive Prototype-based Vision Transformers (NeurIPS 2024 · similarity 0.86)
- LucidPPN: Unambiguous Prototypical Parts Network for User-centric Interpretable Computer Vision (ICLR 2025 · similarity 0.86)
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.
