Reinforcement Learning: A Tutorial

演讲人: 章宗长教授 苏州大学
时间: 2016-01-15 10:00-2016-01-15 12:00
地点:FIT 1-222

 Reinforcement learning (RL), as a branch of machine learning concerned with the ways of optimizing agents’ control of an environment, has made exciting advances in both theory and practice that are increasing its applicability to real-life problems. In this talk, we start with key challenges in RL and existing methods for addressing them, including action selection, temporal difference learning, and function approximation. Then we introduce the principles of representative model-based and model-free RL approaches. Finally, we discuss the RL-related results in a few well-studied games: Backgammon, Chess, Tetris, Go, and 49 arcade video games.