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 2306.09554cs.LG
收藏

通过在线敏感性采样实现探索的低切换策略梯度

Low-Switching Policy Gradient with Exploration via Online Sensitivity Sampling

Microsoft Research(微软研究院);University of California, Los Angeles(加利福尼亚大学洛杉矶分校)

Yunfan Li, Yiran Wang, Yu Cheng, Lin Yang

AI总结:

针对强化学习策略优化方法收敛慢、样本效率低且现有成果难推广到非线性策略的问题,结合基于值的算法技术提出低切换样本高效的LPO算法,大幅提升样本界并经深度网络实证验证。

摘要:

策略优化方法是强化学习(RL)中功能强大的算法,因其能灵活处理策略参数化,且具备应对模型误设的能力。然而,这类方法通常存在收敛速度慢、样本复杂度差的问题。因此,设计具备可证明样本效率的策略优化算法至关重要。但目前该问题的最新研究进展仅在表格型和线性设置下取得成功,这类设置的良好结构无法推广到非线性参数化策略。\n本文针对这一问题,借助基于值的算法的最新研究成果(包括有界eluder维度和在线敏感性采样),设计了一种适用于通用非线性函数逼近的低切换、样本高效的策略优化算法LPO。我们证明,该算法仅需$\u007fO(\frac{\text{poly}(d)}{\u03b5^3})$个样本即可获得$\u03b5$-最优策略,其中$\u03b5$为次优间隙,$d$是逼近策略的函数类的复杂度度量。这一结果大幅优化了此前策略优化算法的最优样本界$\u007fO(\frac{\text{poly}(d)}{\u03b5^8})$。此外,我们使用深度神经网络对理论进行了实证测试,验证了该理论思路的优势。

英文摘要:

Policy optimization methods are powerful algorithms in Reinforcement Learning (RL) for their flexibility to deal with policy parameterization and ability to handle model misspecification. However, these methods usually suffer from slow convergence rates and poor sample complexity. Hence it is important to design provably sample efficient algorithms for policy optimization. Yet, recent advances for this problems have only been successful in tabular and linear setting, whose benign structures cannot be generalized to non-linearly parameterized policies. In this paper, we address this problem by leveraging recent advances in value-based algorithms, including bounded eluder-dimension and online sensitivity sampling, to design a low-switching sample-efficient policy optimization algorithm, LPO, with general non-linear function approximation. We show that, our algorithm obtains an $\varepsilon$-optimal policy with only $\widetilde{O}(\frac{\text{poly}(d)}{\varepsilon^3})$ samples, where $\varepsilon$ is the suboptimality gap and $d$ is a complexity measure of the function class approximating the policy. This drastically improves previously best-known sample bound for policy optimization algorithms, $\widetilde{O}(\frac{\text{poly}(d)}{\varepsilon^8})$. Moreover, we empirically test our theory with deep neural nets to show the benefits of the theoretical inspiration.

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。

↑