Reinforcement Learning

Actor-Critic Algorithms

Actor-Critic Algorithms Math Analysis Methods that learn approximations to both policy and value functions are often called actor–critic methods, where ‘actor’ is a reference to the learned policy, and ‘critic’ refers to the learned value function, usually a state-value function. 这篇论文提出和分析了一类actor–critic算法,用于一参数化系列的随机平稳策略的马尔可夫决策过程(MDP)的基于仿真的优化。 Critic : 一个线性近似架构的TD学习, value function。 Actor : 使用Critic提供的信息,在一个近似梯度方向更新。决策, policy。 大多数强化学习和神经动态编程方法主要属于以下两类中的一类: (a)Actor-only,一系列参数化的策略。性能梯度,对于actor参数的偏导,在提高方向上更新参数。可能的缺点是梯度估计者可能有大偏差,而且,策略改变后,新策略估计与过去估计无关,所以,没有“学习”,也就是没有积累和固化老信息。 (b)Critic-only,只依赖近似值函数,目的是学习Bellman公式的近似解,即希望规定一个近-优化的策略。这个方法不是在策略空间直接优化。这个方法可能可以找到值函数的“好的“近似函数,但是在结果策略的近-优化方面缺乏可信度。 Actor–critic方法结合了actor-only和critic-only的优点。Critic使用一个近似架构和仿真学习值函数,然后用来在性能提高方向上更新actor策略参数,这个方法是基于梯度,可以得到希望的收敛特性,critic-only只有在非常有限的设置才能保证收敛。对比actor-only方法,这种方法能更快地收敛,因为偏差降低了。 这篇论文提出actor-critic… read more »

Policy Gradient Methods for Reinforcement Learning with Function Approximation

  Policy Gradient Methods for Reinforcement Learning with Function Approximation Math Analysis Markov Decision Processes and Policy Gradient So far in this book almost all the methods have been action-value methods; they learned the values of actions and then selected actions based on their estimated action values; their policies would not even exist without the… read more »

Metric spaces

Metric spaces 度量空间 及相关的一些知识点 Definition 6.1.5 (Convergence of sequences). Let ε > 0 be a real number, and let L be a real number. A sequence  of real numbers is said to be ε-close to L iff an is ε-close to L for every n ≥ N, i.e., we have |an − L| ≤ ε… read more »

Reinforcement Learning with Soft State Aggregation

Reinforcement Learning with Soft State Aggregation Math Analysis – Present A New Approach Based On Bayes’ Theorem: Apply Clustering π Rather than State Lookup Table for Computing Q Value Problem Definition and Summary of Notation We consider the problem of solving large Markovian decision processes (MDPs) using RL algorithms and compact function approximation. The objective is to maximize… read more »

Sidebar



×

Google Scholar