Playing the game of Congklak with reinforcement learning

Reinforcement learning is a branch of machine learning that allows an agent to learn to take an action based on its observations and rewards it obtains. In this paper, reinforcement learning agents are trained to play the game of Congklak, a traditional game from Indonesia and Malaysia. Congklak is a deterministic board game played by 2 players which play in turns. However, it was found that the common rules of Congklak make it possible for the first player to win the game without giving a turn to the second player. A change in rule is suggested to make the game more fair and more challenging to train artificial intelligent agents to play the game. The agents were trained based on the suggested rules using model-free reinforcement learning method combined with artificial neural network. After being trained in 3000 games against a random-moves opponent, the agents successfully beat the opponent with a winning chance of up to 90% without any prior knowledge of the game, not even the rules of the game.