# Reinforcement Learning

Data science course for Bachelor econometric students

Coure Description on UvA’s website

This is an introductory-level course in reinforcement learning, with a focus on computational aspects of tabular solution methods, i.e. problems in which the state an action spaces are small enough for approximate value functions to be represented as arrays and tables. Applications of multi-agent models are discussed within an economic context.

## Topics:

- Multi-armed bandits and algorithms;
- Finite Markov Decision processes;
- Dynamic programming;
- Temporal-difference learning;
- N-step bootstrapping;
- Planning and learning with tabular methods;
- Bell-equations, exact and approximate solutions;
- Multi-agent models, games.
- The programming language used is Python/R.

## Study materials

- R. Sutton and A. Barto (2018).
*Reinforcement Learning*, second edition, Cambridge, MA: The MIT Press, ISBN 9780262039246, ca. 200 pages. Free PDF book available at the authors’ homepage: Textbook PDF - Reader Introduction to Python;
- Y. Shoham and K. Leyton-Brown. Multi-agent Systems. Online available at http://www.masfoundations.org.