arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2007.10229cs.LGstat.ML
收藏

一种多臂强盗马尔可夫策略梯度算法的后悔分析

Regret Analysis of a Markov Policy Gradient Algorithm for Multi-arm Bandits

University of Manchester(曼彻斯特大学)

Denis Denisov, Neil Walton

摘要:

本文考虑了一个求解有限臂强盗问题的策略梯度算法。我们允许学习率取决于算法的当前状态,而不是使用确定性的时间递减的学习率。算法的状态在概率单纯形上形成一个马尔可夫链。我们应用福斯特-李雅普诺夫技术来分析这个马尔可夫链的稳定性。我们证明了,如果学习率是很好的选择,那么策略梯度算法是一个瞬态马尔可夫链和链的状态收敛于最佳臂对数或多对数遗憾。

英文摘要:

We consider a policy gradient algorithm applied to a finite-arm bandit problem with Bernoulli rewards. We allow learning rates to depend on the current state of the algorithm, rather than use a deterministic time-decreasing learning rate. The state of the algorithm forms a Markov chain on the probability simplex. We apply Foster-Lyapunov techniques to analyse the stability of this Markov chain. We prove that if learning rates are well chosen then the policy gradient algorithm is a transient Markov chain and the state of the chain converges on the optimal arm with logarithmic or poly-logarithmic regret.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑