Fast-rate PAC-Bayes Generalization Bounds via Shifted Rademacher Processes
arXiv:1908.07585
Abstract
The developments of Rademacher complexity and PAC-Bayesian theory have been largely independent. One exception is the PAC-Bayes theorem of Kakade, Sridharan, and Tewari (2008), which is established via Rademacher complexity theory by viewing Gibbs classifiers as linear operators. The goal of this paper is to extend this bridge between Rademacher complexity and state-of-the-art PAC-Bayesian theory. We first demonstrate that one can match the fast rate of Catoni's PAC-Bayes bounds (Catoni, 2007) using shifted Rademacher processes (Wegkamp, 2003; Lecué and Mitchell, 2012; Zhivotovskiy and Hanneke, 2018). We then derive a new fast-rate PAC-Bayes bound in terms of the "flatness" of the empirical risk surface on which the posterior concentrates. Our analysis establishes a new framework for deriving fast-rate PAC-Bayes bounds and yields new insights on PAC-Bayesian theory.
arXiv comment18 pages
Cited by in corpus (5)
- User-friendly introduction to PAC-Bayes bounds
- NeurIPS 2020 Competition: Predicting Generalization in Deep Learning
- PAC-Bayes, MAC-Bayes and Conditional Mutual Information: Fast rate bounds that handle general VC classes
- De-randomized PAC-Bayes Margin Bounds: Applications to Non-convex and Non-smooth Predictors
- Towards Understanding Generalization via Decomposing Excess Risk Dynamics