arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2001.00805cs.LGcs.AI
收藏

理解强化学习和概率推理

Making Sense of Reinforcement Learning and Probabilistic Inference

DeepMind(深度思维)

Brendan O'Donoghue, Ian Osband, Catalin Ionescu

摘要:

强化学习(RL)将控制问题与统计估计相结合:系统动态对智能体来说是未知的,但可以通过经验来学习。最近的一系列研究将RL视为推理,并提出了一个特定的框架来将RL问题概括为概率推理。我们的论文表面在这种方法的一个关键缺点,并澄清了RL可以连贯地铸造作为一个推理问题的意义。特别是,RL代理必须考虑其行为对未来奖励和观察的影响:探索-开发权衡。除了最简单的设置,在所有情况下,由此产生的推理在计算上是难以处理的,因此实际的RL算法必须采用近似。我们证明了流行的“RL作为推理”近似在非常基本的问题中也会表现不佳。然而,我们表明,通过一个小的修改,框架确实产生了可以证明表现良好的算法,并且我们表明,由此产生的算法相当于最近提出的K-学习,我们进一步将其与汤普森采样联系起来。

英文摘要:

Reinforcement learning (RL) combines a control problem with statistical estimation: The system dynamics are not known to the agent, but can be learned through experience. A recent line of research casts `RL as inference' and suggests a particular framework to generalize the RL problem as probabilistic inference. Our paper surfaces a key shortcoming in that approach, and clarifies the sense in which RL can be coherently cast as an inference problem. In particular, an RL agent must consider the effects of its actions upon future rewards and observations: The exploration-exploitation tradeoff. In all but the most simple settings, the resulting inference is computationally intractable so that practical RL algorithms must resort to approximation. We demonstrate that the popular `RL as inference' approximation can perform poorly in even very basic problems. However, we show that with a small modification the framework does yield algorithms that can provably perform well, and we show that the resulting algorithm is equivalent to the recently proposed K-learning, which we further connect with Thompson sampling.

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。

↑