Regret Analysis of the Anytime Optimally Confident UCB Algorithm
Machine Learning
2016-05-09 v2 Statistics Theory
Machine Learning
Statistics Theory
Abstract
I introduce and analyse an anytime version of the Optimally Confident UCB (OCUCB) algorithm designed for minimising the cumulative regret in finite-armed stochastic bandits with subgaussian noise. The new algorithm is simple, intuitive (in hindsight) and comes with the strongest finite-time regret guarantees for a horizon-free algorithm so far. I also show a finite-time lower bound that nearly matches the upper bound.
Keywords
Cite
@article{arxiv.1603.08661,
title = {Regret Analysis of the Anytime Optimally Confident UCB Algorithm},
author = {Tor Lattimore},
journal= {arXiv preprint arXiv:1603.08661},
year = {2016}
}
Comments
16 pages