Research map / Label-efficient and robust learning / Recognition and long-tail
ProtoVAE: A Trustworthy Self-Explainable Prototypical Variational Model
Srishti Gautam, Ahcene Boubekki, Stine Hansen, Suaiba Amina Salahuddin, Robert Jenssen, Marina MC Höhne, Michael Kampffmeyer
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 |
|---|---|---|---|
| Use ProtoVAE instead of the black-box encoder | ProtoVAE · MNIST, fMNIST, CIFAR-10, QuickDraw, SVHN | accuracy: MNIST: 99.2$\pm$0.1 -> 99.4$\pm$0.1; fMNIST: 91.5$\pm$0.2 -> 91.9$\pm$0.2; CIFAR-10: 83.9$\pm$0.1 -> 84.6$\pm$0.1; Quick | Mixed |
| Add the orthogonality loss (ablation) | ProtoVAE · MNIST | accuracy: 99.2$\pm$0.1 -> 99.4$\pm$0.1 | Improved |
| Add the KL loss (ablation) | ProtoVAE · MNIST | accuracy: 58.8$\pm$10.7 -> 99.4$\pm$0.1 | Improved |
| Add both orthogonality and KL losses (ablation) | ProtoVAE · MNIST | accuracy: 98.9$\pm$0.1 -> 99.4$\pm$0.1 | Improved |
| Use ProtoVAE instead of ProtoPNet | Self-explainable image classifiers · MNIST, fMNIST, CIFAR-10, QuickDraw, SVHN | accuracy: MNIST: 94.7$\pm$0.6 -> 99.4$\pm$0.1; fMNIST: 85.4$\pm$0.6 -> 91.9$\pm$0.2; CIFAR-10: 67.8$\pm$0.9 -> 84.6$\pm$0.1; Quick | Improved |
Closest papers by meaning
- Why Prototypes Collapse: Diagnosing and Preventing Partial Collapse in Prototypical Self-Supervised Learning (ICLR 2026 · similarity 0.87)
- Pantypes: Diverse Representatives for Self-Explainable Models (AAAI 2024 · similarity 0.86)
- Learning to Select Prototypical Parts for Interpretable Sequential Data Modeling (AAAI 2023 · similarity 0.86)
- MCPNet: An Interpretable Classifier via Multi-Level Concept Prototypes (CVPR 2024 · similarity 0.84)
- ProtoArgNet: Interpretable Image Classification with Super-Prototypes and Argumentation (AAAI 2025 · similarity 0.84)
- Projecting Assumptions: The Duality Between Sparse Autoencoders and Concept Geometry (NeurIPS 2025 · similarity 0.84)
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.
