Refined PAC-Bayes Bounds for Offline Bandits
Not on the map. Why?
Abstract (arXiv)
In this paper, we present refined probabilistic bounds on empirical reward estimates for off-policy learning in bandit problems. We build on the PAC-Bayesian bounds from Seldin et al. (2010) and improve on their results using a new parameter optimization approach introduced by Rodr'iguez et al. (2024). This technique is based on a discretization of the space of possible events to optimize the "in probability" parameter. We provide two parameter-free PAC-Bayes bounds, one based on Hoeffding-Azuma's inequality and the other based on Bernstein's inequality. We prove that our bounds are almost optimal as they recover the same rate as would be obtained by setting the "in probability" parameter after the realization of the data.
Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 7 of its 26 references that are in the map. First posted on arXiv 17 Feb 2025.
Builds on
7 of 26 references.
References · 7
- 2017Proximal Policy Optimization Algorithmscited by 14,441
- 2015Trust Region Policy Optimizationcited by 2,503
- 2015Counterfactual Risk Minimization: Learning from Logged Bandit Feedbackcited by 33
- 2023More PAC-Bayes bounds: From bounded losses, to losses with general tail behaviors, to anytime validitycited by 17
- 2021Information Complexity and Generalization Boundscited by 14
- 2024An Information-Theoretic Approach to Generalization Theorycited by 9
- 2022PAC-Bayes Bounds for Bandit Problems: A Survey and Experimental Comparisoncited by 7
Built on by
Not cited yet.