Multi-Agent Reinforcement Learning: Chapter 8 Deep Reinforcement Learning

Last meeting we concluded Chapter 6 of Multi-Agent Reinforcement Learning: Foundations and Modern Approaches and Part 1 of the book as a whole which focuses on so called "Tabular Problems". These problems are characterized by having a state space small enough that we can attempt to estimate values and policies for each state independently as a distinct value.

Part 2 of the book focuses on problems for which this is not possible because the state space is either infinite or so large that it is impractical to track individual states. In order to tackle these problems, we must use a form of function approximation that can generalize values across an arbitrary state space despite having a well defined structure with a finite number of parameters. In the current era, deep neural networks are the method of choice for function approximation in general as well as in reinforcement learning. We will introduce the general method of optimizing neural network function approximators and see the first examples of how they are used in reinforcement learning and multi-agent problems.

As usual you can find below links to the textbook, previous chapter notes, slides, and recordings of some of the previous meetings.

Meetup Links:
Recordings of Previous RL Meetings
Recordings of Previous MARL Meetings
Short RL Tutorials
My exercise solutions and chapter notes for Sutton-Barto
My MARL repository
Kickoff Slides which contain other links
MARL Kickoff Slides

MARL Links:
Multi-Agent Reinforcement Learning: Foundations and Modern Approaches
MARL Summer Course Videos
MARL Slides

Sutton and Barto Links:
Reinforcement Learning: An Introduction by Richard S. Sutton and Andrew G. Barto
Video lectures from a similar course