AI Agents for Gomoku

[ZhenxiangWang]

Gomoku, also called Gobang or Five in a Row, is an abstract strategy board game. It is traditionally played with Go pieces (black and white stones) on a Go board, using 15×15 of the 19×19 grid intersections. Players alternate turns placing a stone of their color on an empty intersection. The winner is the first player to form an unbroken chain of five stones horizontally, vertically, or diagonally.

We want to start researching from the following questions:

1. Can Alpha Zero's algorithm be applied well to gomoku?
2. Can AI agent using Alpha Zero's algorithm discover some remarkable Gomoku game knowledge during its self-play training process?
3. In Gomoku, can Alpha Zero's algorithm perform better than other algorithms?
4. After comparing the performance of multiple algorithms, can we make a better algorithm (in the future)?

Methods:

1. Deep Reinforcement Learning + Monte Carlo Tree Search (Alpha Zero)
2. Alpha–beta Pruning
3. Adaptive Dynamic Programming
4. Hand-coded rules

Baseline:
Some naive strategies used by human beginners.

Related work:
Silver, D., Schrittwieser, J., Simonyan, K., Antonoglou, I., Huang, A., Guez, A., ... & Chen, Y. (2017). Mastering the game of go without human knowledge. Nature, 550(7676), 354.

Zhao, D., Zhang, Z., & Dai, Y. (2012). Self-teaching adaptive dynamic programming for Gomoku. Neurocomputing, 78(1), 23-29.

Shao, K., Zhao, D., Tang, Z., & Zhu, Y. (2016, November). Move prediction in Gomoku using deep learning. In Chinese Association of Automation (YAC), Youth Academic Annual Conference of (pp. 292-297). IEEE.

Tan, Q., & Hu, X. CS221 Project Final Report Gomoku Game Agent.