Research map / Probabilistic and scientific modeling / Variational and generative sampling
Improving Generative Moment Matching Networks with Distribution Partition
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 distribution partitioning with the GMMN-DP generator instead of vanilla GMMN | GMMN · CIFAR-10 | Inception Score: 3.47 ± 0.03 -> 7.49 ± 0.06 | Helped |
| Use distribution partitioning with kernel bandwidth h = 0.08 instead of vanilla GMMN | GMMN · CIFAR-10 | Inception Score: 3.47 ± 0.03 -> 6.21 ± 0.06 | Helped |
| Use distribution partitioning with kernel bandwidth h = 0.15 instead of vanilla GMMN | GMMN · CIFAR-10 | Inception Score: 3.47 ± 0.03 -> 5.08 ± 0.06 | Helped |
| Increase KDE bandwidth from h = 0 to h = 0.08 (ablation) | GMMN-DP · CIFAR-10 | Inception Score: 7.49 ± 0.06 -> 6.21 ± 0.06 | Hurt |
| Increase KDE bandwidth from h = 0.08 to h = 0.15 (ablation) | GMMN-DP · CIFAR-10 | Inception Score: 6.21 ± 0.06 -> 5.08 ± 0.06 | Hurt |
| Use a VAE-pretrained mapping instead of a random mapping (ablation) | GMMN-DP · CIFAR-10 | convergence speed | Helped |
Closest papers by meaning
- Exact Gaussian Moment Matching for Residual Networks: a Second-Order Method (NeurIPS 2026 · similarity 0.85)
- Inductive Moment Matching (ICML 2025 · similarity 0.84)
- Energy-based generator matching: A neural sampler for general state space (NeurIPS 2025 · similarity 0.83)
- M3D: Dataset Condensation by Minimizing Maximum Mean Discrepancy (AAAI 2024 · similarity 0.83)
- Generative Marginalization Models (ICML 2024 · similarity 0.82)
- Generator Matching: Generative modeling with arbitrary Markov processes (ICLR 2025 · 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.