Programmed Checkers AI in C++ using Monte Carlo Tree Search for CS 171A
Variations using different heuristics, ultimately the bare Monte Carlo Tree Search had the best performance. Tested performance using a combination of C and bash scripts to run the agents against random AI's in parallel, batches of 100. Winrate was determined by this.
Winrate vs Random AI 100%. Vs medium level AI has around 96% win rate.