An Analysis of Ensemble Sampling

Authors: Chao Qin, Zheng Wen, Xiuyuan Lu, Benjamin Van Roy

NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical In this paper, we establish a regret bound that ensures desirable behavior when ensemble sampling is applied to the linear bandit problem. This represents the first rigorous regret analysis of ensemble sampling and is made possible by leveraging information-theoretic concepts and novel analytic techniques that may prove useful beyond the scope of this paper. We offer in this paper the first rigorous regret analysis of ES. Like Lu and Van Roy [2017], we study ES applied to linear-Gaussian bandits. This serves as a simple sanity check for the approach. We establish a Bayesian regret bound (Theorem 1) that consists of two terms.
Researcher Affiliation Collaboration Chao Qin Columbia University cqin22@gsb.columbia.edu Zheng Wen Xiuyuan Lu Benjamin Van Roy Deep Mind {zhengwen,lxlu,benvanroy}@google.com
Pseudocode Yes Algorithm 1 Ensemble Sampling
Open Source Code No This paper does not include any experimental results.
Open Datasets No This paper does not include any experimental results.
Dataset Splits No This paper does not include any experimental results.
Hardware Specification No This paper does not include any experimental results.
Software Dependencies No This paper does not include any experimental results.
Experiment Setup No This paper does not include any experimental results.