Research map / Probabilistic and scientific modeling / Variational and generative sampling
Infinite Gaussian Mixture Modeling with an Improved Estimation of the Number of Clusters
Avi Matza, Yuval Bistritz
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 |
|---|---|---|---|
| Replace the baseline concentration-parameter prior with an informative inverse χ2 prior and modified Gibbs sampling | IGMM · p1 synthetic | estimated components (K): 23 (21) -> 6 (94) | Improved |
| Replace the baseline concentration-parameter prior with an informative inverse χ2 prior and modified Gibbs sampling | IGMM · p2 synthetic | estimated components (K): 64 (38) -> 3 (84) | Improved |
| Replace the baseline concentration-parameter prior with an informative inverse χ2 prior and modified Gibbs sampling | IGMM · p3 synthetic | estimated components (K): 43 (32) -> 20 (62) | Improved |
| Replace the baseline concentration-parameter prior with an informative inverse χ2 prior and modified Gibbs sampling | IGMM · p4 synthetic | estimated components (K): 63 (36) -> 30 (61) | Improved |
| Use an inverse χ2 prior for α with the modified training procedure | IGMM · p1 synthetic | symmetric KL divergence: 0.027 -> 0.003 | Improved |
| Use an inverse χ2 prior for α with the modified training procedure | IGMM · p2 synthetic | symmetric KL divergence: 0.082 -> 0.041 | Improved |
| Use an inverse χ2 prior for α with the modified training procedure | IGMM · p3 synthetic | symmetric KL divergence: 13.37 -> 13.51 | Hurt |
| Use an inverse χ2 prior for α with the modified training procedure | IGMM · p4 synthetic | symmetric KL divergence: 39.95 -> 36.28 | Improved |
| Calibrate θ roughly to the anticipated cluster-count range (ablation) | Modified IGMM · p1–p4 synthetic | estimated components (K): θ = 14, 16; θ = 18; θ = 20, 22, 24, 26; θ around 30 (±6); θ ranging from 12 to 22; K = 21 | Mixed |
| Replace secondary adaptive rejection sampling with closed-form posterior sampling | Modified IGMM · Four synthetic datasets | runtime (seconds per dataset): 996 seconds -> 223 seconds | Improved |
| Replace the baseline IGMM training procedure with the modified IGMM procedure | IGMM · Galaxy | estimated components (K): 52 -> 4 | Improved |
| Replace the baseline IGMM training procedure with the modified IGMM procedure | IGMM · Enzyme | estimated components (K): 54 -> 5 | Improved |
Closest papers by meaning
- On the Estimation of Gaussian Mixture Copula Models (ICML 2023 · similarity 0.87)
- Local Maxima in the Likelihood of Gaussian Mixture Models: Structural Results and Algorithmic Consequences (NeurIPS 2016 · similarity 0.86)
- Toward Global Convergence of Gradient EM for Over-Paramterized Gaussian Mixture Models (NeurIPS 2024 · similarity 0.86)
- Learning Mixtures of Gaussian Processes through Random Projection (ICML 2024 · similarity 0.85)
- Achieving Optimal Clustering in Gaussian Mixture Models with Anisotropic Covariance Structures (NeurIPS 2024 · similarity 0.84)
- Is $\sqrt{d}$ Separation Necessary for Gradient EM to Learn Gaussian Mixtures in High Dimensions? (NeurIPS 2026 · 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.
