A KL-LUCB Bandit Algorithm for Large-Scale Crowdsourcing
Statistics Theory
2017-09-13 v1 Statistics Theory
Abstract
This paper focuses on best-arm identification in multi-armed bandits with bounded rewards. We develop an algorithm that is a fusion of lil-UCB and KL-LUCB, offering the best qualities of the two algorithms in one method. This is achieved by proving a novel anytime confidence bound for the mean of bounded distributions, which is the analogue of the LIL-type bounds recently developed for sub-Gaussian distributions. We corroborate our theoretical results with numerical experiments based on the New Yorker Cartoon Caption Contest.
Cite
@article{arxiv.1709.03570,
title = {A KL-LUCB Bandit Algorithm for Large-Scale Crowdsourcing},
author = {Bob Mankoff and Robert Nowak and Ervin Tanczos},
journal= {arXiv preprint arXiv:1709.03570},
year = {2017}
}