150 likes | 255 Vues
This presentation provides an introduction to Reinforcement Learning (RL), covering its fundamental concepts and approaches. It discusses how agents interact with environments to maximize cumulative rewards over time using observations and actions. The key ideas of Markov Decision Processes (MDP), temporal differences, and functional approximation are explored. The talk draws from various sources, including lecture videos by noted experts and the book "Introduction to Reinforcement Learning" by Sutton and Barto. This is an essential resource for those interested in machine learning and sequential decision-making.
E N D
Introduction to Reinforcement Learning HirenAdesara Prof: Dr. Gittens
Sources for this presentation • Lecture videos of • Mr. Satinder Singh, University of Michigan. • Douglas Aberdeen, Australian National University. From www.videolectures.net • Book : Introduction to Reinforcement Learning by Sutton and Barto (http://www.cs.ualberta.ca/%7Esutton/book/ebook/the-book.html)
Another View of RL • Observation-Action-Response. • O1a1r1o2a2r2o3a3r3 • Agent chooses action so as to maximize expected cumulative reward over time. • Observations can be vectors or other structures. • Actions are multi-dimensional. • Rewards are scalar. (known or unknown). • Agents have partial knowledge about environment.
RL and Machine Learning • Supervised Learning • Learning approach to regression and classification. • Learning from example and learning from teacher. • Unsupervised learning • Learning approaches to dimensionality reduction, density estimation and recording data based on some principles. • Reinforcement Learning • Learning approaches to sequential decision making. • Learning from critics, learning from delayed reward.
Key ideas of RL • Markov Decision Process(MDP). • Temporal Differences( updating a guess on the basis of the previous guess). • Functional approximation.