A brief tutorial on reinforcement learning: The game of Chung Toi.
Christopher J. Gatti, Jonathan D. Linton, Mark J. Embrechts · 2011
Abstract. This work presents a simple implementation of reinforcement learning, using the temporal difference algorithm and a neural network, applied to the board game of Chung Toi, which is a challenging variation of Tic-Tac-Toe. The implementation of this learning algorithm is fully described and includes all parameter settings and various techniques to improve the ability of the network to learn the board game. With relatively little training, the network was able to win nearly 90 % of games played against a ’smart ’ random opponent. The aim of this work is to develop a general software framework for reinforcement learning with an aim to allow for the implementation of game playing strategies for managers that can be applied to option and portfolio management. 1