Project „Board Game AI 2“

Keywords: Reinforcement Learning, Program Synthesis, Monte Carlo Tree Search

Project Context and Goal: The objective of this international team project is the development of an intelligent agent to play the board game Turing Tumble. The game agent should be trained using reinforcement learning (RL) in order to solve simple logic challenges building on the implementation of a predecessor project. Thus, the existing approach can be used to get familiar with reinforcement learning and should subsequently be extended. In particular, the RL Agent should be enhanced to make use of Monte Carlo Tree Search. In addition, a UI should be implemented to illustrate the solutions of the agent.