Research map / Probabilistic and scientific modeling / Variational and generative sampling
Learning Energy-Based Model with Variational Auto-Encoder as Amortized Sampler
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 ancestral sampling from a VAE to initialize finite-step Langevin dynamics | EBM image generation · CIFAR-10 | Inception score: 6.65 vs 6.55 | Helped |
| Use ancestral Langevin sampling and variational MCMC teaching | image completion · Facade | PSNR: 20.47 -> 21.62 | Helped |
| Use ancestral Langevin sampling and variational MCMC teaching | image completion · StreetView | PSNR: 21.17 -> 22.61 | Helped |
| Increase Langevin steps and reduce step size (ablation) | ancestral Langevin sampling · CIFAR-10 | Inception score: 3.606 to 6.143; 5.146 to 3.482 | Mixed |
| Set variational loss penalty γ near 2 (ablation) | variational MCMC teaching · CIFAR-10 | Inception score: 5.106 -> 6.159; γ=10: 4.693 | Helped |
| Vary latent dimension (ablation) | ancestral Langevin sampling · CIFAR-10 | Inception score: 5.973 to 6.213 | No clear effect |
Closest papers by meaning
- Learning Energy-based Model via Dual-MCMC Teaching (NeurIPS 2023 · similarity 0.91)
- No MCMC for me: Amortized sampling for fast and stable training of energy-based models (ICLR 2021 · similarity 0.91)
- Energy-Inspired Models: Learning with Sampler-Induced Distributions (NeurIPS 2019 · similarity 0.91)
- Learning Energy-Based Models by Cooperative Diffusion Recovery Likelihood (ICLR 2024 · similarity 0.88)
- Online Variational Sequential Monte Carlo (ICML 2024 · similarity 0.87)
- Joint Learning of Energy-based Models and their Partition Function (ICML 2025 · similarity 0.87)
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.