arXivDaily arXiv每日学术速递 周一至周五更新
北美退伍军人节(Veterans Day),12 日无更新
arXiv 2210.00701cs.LGcs.AIstat.ML
收藏

线性函数逼近下无奖励强化学习的近优部署效率

Near-Optimal Deployment Efficiency in Reward-Free Reinforcement Learning with Linear Function Approximation

UC Santa Barbara(加州大学圣巴巴拉分校)

Dan Qiao, Yu-Xiang Wang

摘要:

我们研究了在\emph{reward-free}探索环境下使用线性函数逼近的部署有效强化学习(RL)问题。这是一个动机很好的问题,因为在现实生活中的RL应用程序中部署新策略是昂贵的。在线性MDP设置与功能维度$d$和规划视野$H$,我们提出了一种新的算法,收集最多$\widetilde{O}(\frac{d^2H^5}{ε^2})$轨迹内$H$部署,以确定$ε$ -最优策略的任何(可能依赖于数据)的选择奖励函数。据我们所知,我们的方法是第一个实现最佳的部署复杂性和最佳的$d$依赖样本复杂性在同一时间,即使奖励是已知的提前。我们的新技术包括探索保持政策离散化和广义G-最优实验设计,这可能是独立的利益。最后,我们分析了低适应RL中遗憾最小化的相关问题,并给出了切换成本和批处理复杂度的信息论下界。

英文摘要:

We study the problem of deployment efficient reinforcement learning (RL) with linear function approximation under the \emph{reward-free} exploration setting. This is a well-motivated problem because deploying new policies is costly in real-life RL applications. Under the linear MDP setting with feature dimension $d$ and planning horizon $H$, we propose a new algorithm that collects at most $\widetilde{O}(\frac{d^2H^5}{ε^2})$ trajectories within $H$ deployments to identify $ε$-optimal policy for any (possibly data-dependent) choice of reward functions. To the best of our knowledge, our approach is the first to achieve optimal deployment complexity and optimal $d$ dependence in sample complexity at the same time, even if the reward is known ahead of time. Our novel techniques include an exploration-preserving policy discretization and a generalized G-optimal experiment design, which could be of independent interest. Lastly, we analyze the related problem of regret minimization in low-adaptive RL and provide information-theoretic lower bounds for switching cost and batch complexity.

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。

↑