arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2006.12484cs.LGcs.AImath.OCstat.ML
收藏

欠完备POMDPs的样本有效强化学习

Sample-Efficient Reinforcement Learning of Undercomplete POMDPs

Princeton University(普林斯顿大学);University of Washington(华盛顿大学);Microsoft Research, NYC(微软研究院纽约分院)

Chi Jin, Sham M. Kakade, Akshay Krishnamurthy, Qinghua Liu

摘要:

部分可观测性是许多强化学习应用中的常见挑战,这需要代理维护记忆,推断潜在状态,并将此过去信息整合到探索中。这一挑战导致了许多计算和统计上的困难结果,用于学习一般的部分可观察马尔可夫决策过程(POMDPs)。这项工作表明,这些硬度障碍并不妨碍有效的强化学习丰富和有趣的子类POMDPs。特别是,我们提出了一个样本有效的算法,OOM-UCB,情节有限欠完备POMDPs,观察的数量大于潜在状态的数量,探索是必不可少的学习,从而区分我们的结果从以前的作品。OOM-UCB实现了用于找到$\varepsilon$最优策略的最优样本复杂度$\tilde{\mathcal{O}}(1/\varepsilon^2)$,以及在所有其他相关量上是多项式的。作为一个有趣的特殊情况下,我们还提供了一个计算和统计上有效的算法POMDPs与确定性的状态转移。

英文摘要:

Partial observability is a common challenge in many reinforcement learning applications, which requires an agent to maintain memory, infer latent states, and integrate this past information into exploration. This challenge leads to a number of computational and statistical hardness results for learning general Partially Observable Markov Decision Processes (POMDPs). This work shows that these hardness barriers do not preclude efficient reinforcement learning for rich and interesting subclasses of POMDPs. In particular, we present a sample-efficient algorithm, OOM-UCB, for episodic finite undercomplete POMDPs, where the number of observations is larger than the number of latent states and where exploration is essential for learning, thus distinguishing our results from prior works. OOM-UCB achieves an optimal sample complexity of $\tilde{\mathcal{O}}(1/\varepsilon^2)$ for finding an $\varepsilon$-optimal policy, along with being polynomial in all other relevant quantities. As an interesting special case, we also provide a computationally and statistically efficient algorithm for POMDPs with deterministic state transitions.

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。

↑