Research map / Learning theory and algorithms / Kernels and regression theory
PAC-Bayesian Bounds on Rate-Efficient Classifiers
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 |
|---|---|---|---|
| Increasing Gaussian input noise variance sigma_c^2 from 0.010 to 1.000 | MinCq majority vote on MNIST binary tasks · MNIST controlled noise | noise invariance bound Bη: 0.149 to 0.470 | Hurt |
| Bounding noise invariance with the PAC-Bayes bound (Theorem 3) under JPEG2000 compression | MinCq majority vote on MNIST-JPEG2000 · MNIST-JPEG2000 q=5 | |ηT_Q − Bη| gap | No clear effect |
| Lowering JPEG2000 quality from q=90 to q=5 (fewer bytes) | MinCq majority vote on MNIST-JPEG2000 · MNIST-JPEG2000 test risk | test risk RT(BQ): 0.24 to 0.28 | Hurt |
Closest papers by meaning
- Learning Stochastic Majority Votes by Minimizing a PAC-Bayes Generalization Bound (NeurIPS 2021 · similarity 0.90)
- Second Order PAC-Bayesian Bounds for the Weighted Majority Vote (NeurIPS 2020 · similarity 0.90)
- On Margins and Generalisation for Voting Classifiers (NeurIPS 2022 · similarity 0.89)
- A New PAC-Bayesian Perspective on Domain Adaptation (ICML 2016 · similarity 0.88)
- Tight Bounds on the Distortion of Randomized and Deterministic Distributed Voting (NeurIPS 2025 · similarity 0.85)
- How many classifiers do we need? (NeurIPS 2024 · similarity 0.84)
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.