Research map / Probabilistic and scientific modeling / Variational and generative sampling
Diversity-Promoting Bayesian Learning of Latent Variable Models
Pengtao Xie, Jun Zhu, Eric Xing
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 a mutual angular prior to BMEM | BMEM on Adult-9 · Adult-9 | accuracy (%): 84.9 -> 88.9 | Improved |
| Add mutual angular posterior regularization | BMEM on Adult-9 · Adult-9 | accuracy (%): 84.9 -> 88.7 | Improved |
| Add a mutual angular prior to BMEM | BMEM on SUN-Building · SUN-Building | accuracy (%): 80.4 -> 85.3 | Improved |
| Add mutual angular posterior regularization | BMEM on SUN-Building · SUN-Building | accuracy (%): 80.4 -> 84.9 | Improved |
| Use an infinite mutual angular prior instead of independent component priors | ILFM on Yale, Block-Images, AR, EEG, and Piano · Yale, Block-Images, AR, EEG, Piano | L2 test error: 447$\pm$7 -> 419$\pm$4; 6.3$\pm$0.4 -> 4.4$\pm$0.2; 926$\pm$4 -> 871$\pm$7; 5382$\pm$34 -> 575$\pm$21; 5.3$\pm$0.1 -> 4. | Improved |
| Use an infinite mutual angular prior instead of independent component priors | ILFM on Reuters, TDT, 20-News, 15-Scenes, and Caltech-101 · Reuters, TDT, 20-News, 15-Scenes, Caltech-101 | clustering accuracy (%): 45.4$\pm$0.3 -> 48.2$\pm$0.6; 48.3$\pm$0.7 -> 53.2$\pm$0.4; 23.7$\pm$0.2 -> 25.2$\pm$0.1; 22.7$\pm$0.2 -> 25.3$\pm$0.2; | Improved |
| Use an infinite mutual angular prior instead of independent component priors | ILFM on Reuters, TDT, 20-News, 15-Scenes, and Caltech-101 · Reuters, TDT, 20-News, 15-Scenes, Caltech-101 | normalized mutual information (%): 41.7$\pm$0.5 -> 45.4$\pm$0.4; 44.2$\pm$0.1 -> 49.6$\pm$0.6; 38.9$\pm$0.8 -> 44.6$\pm$0.9; 42.1$\pm$0.7 -> 47.5$\pm$0.2; | Improved |
| Use an infinite mutual angular prior instead of independent component priors | ILFM on Reuters · Reuters infrequent categories 3–9 | precision@100 (%): 7.5 -> 36.2; 8.3 -> 37.8; 6.9 -> 29.1; 7.2 -> 20.4; 2.6 -> 8.3; 3.8 -> 13.8; 3.4 -> 11.6 | Improved |
| Use type II MABN with variational inference rather than type I MABN with MCMC | BMEM on Adult-9 · Adult-9 | training time (hours): 7.5 -> 2.9 | Improved |
Closest papers by meaning
- Forward $\chi^2$ Divergence Based Variational Importance Sampling (ICLR 2024 · similarity 0.86)
- Variational Inference with Mixtures of Isotropic Gaussians (NeurIPS 2025 · similarity 0.86)
- Environment-Robust Representation Learning with Empirical Bayes (NeurIPS 2026 · similarity 0.85)
- Efficient Mixture Learning in Black-Box Variational Inference (ICML 2024 · similarity 0.85)
- Variational Learning of Fractional Posteriors (ICML 2025 · similarity 0.85)
- Learning Latent Variable Models via Jarzynski-adjusted Langevin Algorithm (NeurIPS 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.
