Reinforcement Learning: Theory and Algorithms

Reinforcement Learning: Theory and AlgorithmsAlekh AgarwalNan JiangSham M. KakadeWen SunJanuary 31, 2022 WORKING DRAFT:Please any typos or errors you appreciate it!iiContents1 Fundamentals31 Markov Decision (Infinite-Horizon) Markov Decision Processes .. objective, policies, and values .. Consistency Equations for Stationary Policies .. Optimality Equations .. Markov Decision Processes .. Complexity .. Iteration .. Iteration .. Iteration for Finite Horizon MDPs .. Linear Programming Approach .. Complexity and Sampling Models .. : Advantages and The Performance Difference Lemma .. Remarks and Further Reading ..202 Sample Complexity with a Generative : a naive model-based approach .. Sample Complexity .. Optimal Sample Complexity (and the Model Based Approach).

A transition function P: SA! ( S), where ( S) is the space of probability distributions over S(i.e., the probability simplex). P(s0js;a) is the probability of transitioning into state s0upon taking action ain state s. We use P s;ato denote the vector P( s;a). A reward function r: SA!

Loading..

Tags:

  Learning, Theory, Probability, Algorithm, Reinforcement, Reinforcement learning, Theory and algorithms

Information