Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Bayesian Combination of Probabilistic Classifiers using Multivariate Normal Mixtures
Gregor Pirš, Erik Štrumbelj
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 |
|---|---|---|---|
| constrain MM covariance matrices to diagonal (MM-diag) (ablation) | MM · Toy, bookies and DNA data sets (Table 1) | log-score: MM-diag performs worse than the MM on all data sets, however the differences are small on average | Hurt |
| add MM predictions as a source to the linear opinion pool (LOP+MM) | LOP · Toy and real-world data sets | log-score: best on average; outperforms LOP on all real-world data sets | Improved |
| add MM predictions as a source to LOP | LOP · Air pollution data sets (Table 2) | log-score: -0.634 -> -0.613 (PM10 tomorrow); -0.419 -> -0.406 (PM10 today); -0.378 -> -0.375 (O3 today); -0.442 -> -0.439 (O3 tomor | Improved |
| use MM (probabilistic multivariate normal mixtures) instead of IBCC | IBCC · DNA data sets | log-score: -0.206 vs -0.248 (DNA A); -0.132 vs -0.181 (DNA B) | Improved |
| use probabilistic inputs via quasi-probabilistic sampling (IBCC QPI) instead of crisp IBCC (ablation) | IBCC · DNA data sets | log-score | Improved |
| use Bayesian combination methods versus best single source on bookies data | MM and IBCC · bookies | log-score: all methods fail to outperform the best source (Source 1, -0.965) | No clear effect |
| add MM to model correlated probabilistic sources in toy B and toy C (continuous relation) | MM · toy B and toy C | log-score: MM -0.893 vs best -0.831 (toy B); -0.828 vs -0.81 (toy C) | Hurt |
| learn latent structure (MM) rather than weighting sources (LOP) on systematic errors | MM · toy D | log-score: -0.835 (LOP) -> -0.001 (MM) | Improved |
Closest papers by meaning
- Data Mixture Optimization: A Multi-fidelity Multi-scale Bayesian Framework (NeurIPS 2025 · similarity 0.83)
- A Provably Accurate Randomized Sampling Algorithm for Logistic Regression (AAAI 2024 · similarity 0.82)
- A Unified View of FANOVA: A Comprehensive Bayesian Framework for Component Selection and Estimation (ICML 2024 · similarity 0.82)
- Conformal Prediction for Ensembles: Improving Efficiency via Score-Based Aggregation (NeurIPS 2025 · similarity 0.82)
- Probabilistic Forecasting of Irregularly Sampled Time Series with Missing Values via Conditional Normalizing Flows (AAAI 2025 · similarity 0.82)
- SimPro: A Simple Probabilistic Framework Towards Realistic Long-Tailed Semi-Supervised Learning (ICML 2024 · similarity 0.82)
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.
