Research map / Probabilistic and scientific modeling / Variational and generative sampling
Amortized Variational Inference for Simple Hierarchical Models
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 |
|---|---|---|---|
| Amortize the local conditional parameters | Dense branch Gaussian · MovieLens25M, 18M ratings | test likelihood: -1.3046e+06 -> -1.1476e+06 | Helped |
| Amortize the local conditional parameters | Block-diagonal branch Gaussian · MovieLens25M, 18M ratings | test likelihood: -1.2538e+06 -> -1.1484e+06 | Helped |
| Amortize the local conditional parameters | Diagonal branch Gaussian · MovieLens25M, 18M ratings | test likelihood: -1.2713e+06 -> -1.1497e+06 | Helped |
| Amortize the local conditional parameters | Dense branch Gaussian · MovieLens25M, 2.5K ratings | test likelihood: -166.66 -> -166.64 | No clear effect |
| Amortize the local conditional parameters | Dense branch Gaussian · MovieLens25M, 180K ratings | test likelihood: -11054.43 -> -10976.38 | Helped |
| Amortize the local conditional parameters | Diagonal branch Gaussian · MovieLens25M, 180K ratings | test likelihood: -10977.95 -> -10980.75 | Hurt |
| Use a block-diagonal rather than dense Gaussian | Amortized branch inference · MovieLens25M, 18M ratings | test likelihood: -1.1476e+06 vs -1.1484e+06 | No clear effect |
| Share local variational parameters through amortization | Branch variational inference · Large hierarchical models | convergence speed | Helped |
Closest papers by meaning
- Variational Inference with Locally Enhanced Bounds for Hierarchical Models (ICML 2022 · similarity 0.93)
- Importance Weighted Hierarchical Variational Inference (NeurIPS 2019 · similarity 0.91)
- Meta-Amortized Variational Inference and Learning (AAAI 2020 · similarity 0.89)
- Provably Scalable Black-Box Variational Inference with Structured Variational Families (ICML 2024 · similarity 0.89)
- Meta-Learning with Shared Amortized Variational Inference (ICML 2020 · similarity 0.89)
- Variational Inference with Mixtures of Isotropic Gaussians (NeurIPS 2025 · similarity 0.88)
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.