Solving Graph-based Public Good Games with Tree Search and Imitation Learning

Neural Information Processing Systems 

Our evaluation results show that this policy is able to reach 99.5%