Research map / Learning theory and algorithms / Kernels and regression theory
Fast-Rate PAC-Bayesian Generalization Bounds for Meta-Learning
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 |
|---|---|---|---|
| Minimize Catoni-bound (Gibbs-style meta objective) instead of kl-bound for classification | meta-learning classification with MLP · Pixel-shuffled MNIST (100/200/300 swaps) | test bound and test error | Helped |
| Minimize kl-bound (fast-rate) meta objective vs. prior PAC-Bayes meta methods | meta-learning classification with MLP · Pixel-shuffled MNIST 100 swaps | test bound | Mixed |
| Gibbs optimal hyper-posterior (Catoni-bound closed form) for regression | regression with Bayesian NN · 5 regression environments (RMSE) | RMSE | Mixed |
| Catoni-bound vs kl-bound in classification convergence (ablation) | meta-learning classification · Pixel-shuffled MNIST convergence | test bound and test error | Helped |
Closest papers by meaning
- Generalization Bounds for Meta-Learning via PAC-Bayes and Uniform Stability (NeurIPS 2021 · similarity 0.94)
- More Flexible PAC-Bayesian Meta-Learning by Learning Learning Algorithms (ICML 2024 · similarity 0.92)
- Fast Rate Bounds for Multi-Task and Meta-Learning with Different Sample Sizes (NeurIPS 2025 · similarity 0.91)
- Generalization Bounds via Meta-Learned Model Representations: PAC-Bayes and Sample Compression Hypernetworks (ICML 2025 · similarity 0.90)
- Learning via Surrogate PAC-Bayes (NeurIPS 2024 · similarity 0.88)
- On the Stability and Generalization of Meta-Learning (NeurIPS 2024 · 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.