Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Averaged Collapsed Variational Bayes Inference
Katsuhiko Ishiguro, Issei Sato, Naonori Ueda
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 |
|---|---|---|---|
| use 0th-order CVB0 approximation instead of 2nd-order CVB (ablation) | CVB on LDA · 20 news group (K=50) | test perplexity: 3394 -> 2009 | Improved |
| add averaging of variational posteriors to CVB0 (ACVB0) (ablation) | CVB0 on LDA · 20 news group (K=50) | test perplexity: 2009 -> 2016 | Hurt |
| add averaging to 2nd-order CVB (ACVB) (ablation) | CVB on LDA · 20 news group (K=50) | test perplexity: 3394 -> 3393 | No clear effect |
| use ACVB0 instead of naive VB (ablation) | VB on LDA · 20 news group (K=50) | test perplexity: 2877 -> 2016 | Improved |
| use ACVB0 instead of naive VB (ablation) | VB on LDA · Enron (K=200) | test perplexity: 2504 -> 1188 | Improved |
| use collapsed Gibbs (1000 iterations) instead of ACVB0 (ablation) | ACVB0 on LDA · 20 news group (K=50) | test perplexity: 2016 -> 1909 | Improved |
| use ACVB0 instead of naive VB (ablation) | VB on IRM · Lastfm UserXUser (K=20) | test log likelihood per entry: -0.0289 -> -0.0270 | Improved |
| use ACVB0 instead of naive VB (ablation) | VB on IRM · Enron Oct. (K=20) | test log likelihood per entry: -0.1274 -> -0.1110 | Improved |
| add averaging to CVB (ACVB) (ablation) | CVB on IRM · Enron Jun. (K=20) | test log likelihood per entry: -0.0559 -> -0.0595 | Hurt |
| use collapsed Gibbs (3000 iterations) instead of ACVB0 (ablation) | ACVB0 on IRM · Enron Oct. (K=20) | test log likelihood per entry: -0.1110 -> -0.1268 | Hurt |
Closest papers by meaning
- Variational Bayes under Model Misspecification (NeurIPS 2019 · similarity 0.90)
- Sparse Variational Inference: Bayesian Coresets from Scratch (NeurIPS 2019 · similarity 0.88)
- Bayesian Online Natural Gradient (BONG) (NeurIPS 2024 · similarity 0.88)
- An Optimization-centric View on Bayes' Rule: Reviewing and Generalizing Variational Inference (JMLR 2022 · similarity 0.87)
- Spike and slab variational Bayes for high dimensional logistic regression (NeurIPS 2020 · similarity 0.87)
- Variational Inference with Mixtures of Isotropic Gaussians (NeurIPS 2025 · 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.
