SeegaAI : Deep Reinforcement Learning in Seega
share
This research paper introduces SeegaAI, a research project to develop a powerful arti cial intelligence bot for the game of Seega using deep reinforcement learning. Researchers have long been working on embedding human-like behavior in computers: a eld of research popularly called arti cial intelligence (hereafter called AI). With the advent of personal, faster and more ecient computers in our present generation, it has become easy to create powerful AIs that can play (and beat humans) at various games like checkers, backgammon, chess, and Go[1]. As a result, several resources concerning the implementation of AI in these games- research papers, python libraries, game bots online, project codes, etc. { are available and easily accessible online. On the contrary, only few resources from African indigenous games (Ayo, Seega, Abula) can be found because little research has been made on them. These games are however unique because they involve a complex combination of strategy, psychology, and quick decision making. For example, playing Seega involves two stages: placing the players and moving the players, which both need strategy to win, unlike chess (which only requires strategy in moving the players as they have a xed position) and Go (which involves only placing the player). It is for this complex gaming structure of Seega that we chose it for our AI project. We hope to achieve two things with this project: successfully implement deep reinforcement learning in Seega and provide research materials for implementing machine learning in African indigenous board games.
READ FULL TEXT