Research map / Vision-language and multimodal / Cross-modal learning
Explainable Mixture Models through Differentiable Rule Learning
Matthias Wilms, Sascha Xu, Jilles Vreeken
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 overlap penalty (lambda = 0.3) (ablation) | XMM-GMM · Real datasets | number of rules: up to 16% fewer rules at λ = 0.3 | Improved |
| add overlap penalty (ablation) | XMM-NSF · Real datasets | number of rules and NLL: rules only decrease significantly at λ = 1 | Mixed |
| use GMM instead of NSF density estimator (ablation) | XMM · UCI real-world datasets | average rank of test NLL: 5.60 -> 4.20 | Improved |
| add BIC model selection (ablation) | XMM-GMM · UCI real-world datasets | average rank of test NLL: 4.20 -> 4.47; rule rank 3.80 -> 1.73 | Mixed |
| add BIC model selection (ablation) | XMM-NSF · UCI real-world datasets | average rank of test NLL: 5.60 -> 6.07 | Hurt |
| k-means++ anchoring instead of random initialization (ablation) | XMM · synthetic data | rule coverage | Improved |
| replace CDTREE with XMM | tree-based CDE · Gold nanoclusters HOMO-LUMO | test NLL: -1.683 -> -1.706 | Improved |
| increase initial number of rules k beyond true count (ablation) | XMM-GMM vs XMM-NSF · synthetic data | excess rules | Mixed |
Closest papers by meaning
- Continuously Parameterized Mixture Models (ICML 2023 · similarity 0.85)
- Explainable Neural Networks with Guarantee: A Sparse Estimation Approach (AAAI 2025 · similarity 0.85)
- Additive Models Explained: A Computational Complexity Approach (NeurIPS 2025 · similarity 0.84)
- Evidential Mixture Machines: Deciphering Multi-Label Correlations for Active Learning Sensitivity (NeurIPS 2024 · similarity 0.84)
- Subtractive Mixture Models via Squaring: Representation and Learning (ICLR 2024 · similarity 0.83)
- End-to-end Learning of Gaussian Mixture Priors for Diffusion Sampler (ICLR 2025 · similarity 0.83)
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.
