/TrainingCell

Reinforcement learning algorithms

Primary LanguageC++MIT LicenseMIT

TrainingCell

Reinforcement learning algorithms

An open source C++ environment to design and train reinforcement learning agents of various types. Based on DeepLearning project https://github.com/dragunovdenis/DeepLearning (uses the latter as a sub-module).

At the moment, the environment provides functionality to design, train and evaluate agents for playing checkers and chess: image image as well as an interface to interactively play with a trained agent image image image .

The project is inspired by the brilliant book "Reinforcement Learning: An Introduction" by Sutton, Richard S. and Barto, Andrew G.