Research map / Learning theory and algorithms / Kernels and regression theory
Meta-Learning PAC-Bayes Priors in Model Averaging
Yimin Huang, Weiran Huang, Liang Li, Zhenguo Li
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 |
|---|---|---|---|
| add sequential batch sampling to RBM | RBM · Example 1 Model 1 (0,5) | MSPE: 48.23 -> 14.08 | Improved |
| add sequential batch sampling to RBM | RBM · Example 1 Model 1 (0,1) | MSPE: 2.03 -> 1.71 | Improved |
| add sequential batch sampling to RBM | RBM · Example 1 Model 3 (0.9,5) | MSPE: 38.51 -> 29.44 | Improved |
| add sequential batch sampling to RBM | RBM · Example 1 Model 1 (0.9,1) | MSPE: 3.71 -> 3.25 | Improved |
| replace SOIL with RBM (risk-bound weights) | SOIL · Example 1 Model 1 (0,5) | MSPE: 53.21 -> 48.23 | Improved |
| replace SOIL with RBM | SOIL · Example 1 Model 1 (0.9,1) | MSPE: 2.17 -> 3.71 | Hurt |
| use historical-data prior (HDR) instead of non-informative prior | RBM · BGS | MSPE: 13.54 -> 13.06 | Improved |
| use historical-data prior (HDR) instead of non-informative prior | RBM · Bardet | MSPE: 0.0054 -> 0.0050 | Improved |
| use historical-data prior (HDR) instead of SOIL | SOIL · BGS | MSPE: 16.74 -> 13.06 | Improved |
| use HDR meta-learned prior instead of MLAP/MAML | MLAP_M · Permuted MNIST 200 swaps | test error (%): 3.4 -> 0.72 | Improved |
| use HDR instead of MAML | MAML · Permuted MNIST 200 swaps | test error (%): 3.77 -> 0.72 | Improved |
| increase number of training tasks | HDR · Permuted MNIST | test error | No clear effect |
Closest papers by meaning
- More Flexible PAC-Bayesian Meta-Learning by Learning Learning Algorithms (ICML 2024 · similarity 0.88)
- Meta-Learning Reliable Priors in the Function Space (NeurIPS 2021 · similarity 0.87)
- Meta-Learning by Adjusting Priors Based on Extended PAC-Bayes Theory (ICML 2018 · similarity 0.87)
- Recursive PAC-Bayes: A Frequentist Approach to Sequential Prior Updates with No Information Loss (NeurIPS 2024 · similarity 0.85)
- Learning via Surrogate PAC-Bayes (NeurIPS 2024 · similarity 0.85)
- Estimating Interventional Distributions with Uncertain Causal Graphs through Meta-Learning (NeurIPS 2025 · similarity 0.83)
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.
