Research map / Probabilistic and scientific modeling / Variational and generative sampling
Learning Multimodal Latent Generative Models with Energy-Based Prior
Shiyu Yuan, Jiali Cui, Hanao Li, Tian Han
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 unimodal Laplacian prior with latent EBM prior | multimodal latent generative model · PolyMNIST | joint coherence: 0.232 -> 0.746 vs MMVAE* | Improved |
| replace unimodal prior with EBM prior | multimodal latent generative model · MNIST-SVHN | joint coherence: 0.215 -> 0.419 vs MMVAE* | Improved |
| EBM prior for cross generation | multimodal latent generative model · PolyMNIST | cross coherence: 0.844 -> 0.853 | No clear effect |
| EBM prior for cross generation on MNIST-SVHN | multimodal latent generative model · MNIST-SVHN S->M | cross coherence: 0.523 -> 0.653 | Improved |
| add EBM prior on top of MMVAE+ modality-specific priors | MMVAE+ · PolyMNIST | joint coherence: 0.344 -> 0.878 | Improved |
| add EBM prior on top of MMVAE+ modality-specific priors | MMVAE+ · PolyMNIST | cross coherence: 0.869 -> 0.897 | Improved |
| increase Langevin MCMC steps from 30 to 50 (ablation) | EBM prior · PolyMNIST | joint coherence: 0.683 -> 0.746 | Improved |
| increase EBM layers from 4 to 6 (D=200) (ablation) | EBM prior · PolyMNIST | joint coherence: 0.574 -> 0.645 | Improved |
| increase EBM layers from 4 to 6 (D=200) (ablation) | EBM prior · PolyMNIST | cross coherence: 0.842 -> 0.832 | No clear effect |
| increase EBM hidden units from 200 to 400 (ablation) | EBM prior · PolyMNIST | joint coherence: 0.574 -> 0.746 | Improved |
Closest papers by meaning
- Efficient Modality Translation via Arbitrary Conditioning and Wasserstein Regularization (AAAI 2026 · similarity 0.89)
- ShaLa: Multimodal Shared Latent Generative Modelling (AAAI 2026 · similarity 0.89)
- Conditional Generative Modeling via Learning the Latent Space (ICLR 2021 · similarity 0.89)
- Learning Latent Space Hierarchical EBM Diffusion Models (ICML 2024 · similarity 0.87)
- Aggregation of Dependent Expert Distributions in Multimodal Variational Autoencoders (ICML 2025 · similarity 0.87)
- Unity by Diversity: Improved Representation Learning for Multimodal VAEs (NeurIPS 2024 · similarity 0.86)
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.
