Monte Carlo Tree Search (MCTS) is a popular approach for tree search in a variety of games. While MCTS allows for fine-grained time control, not much has been published on time management for MCTS programs under tournament conditions. This paper first investigates the effects of various time-management strategies on playing strength in the challenging game of Go. A number of domain-independent strategies are then tested in the domains Connect-4, Breakthrough, Othello, and Catch the Lion. We consider strategies taken from the literature as well as newly proposed and improved ones. Strategies include both semi-dynamic strategies that decide about time allocation for each search before it is started, and dynamic strategies that influence the duration of each move search while it is already running. Furthermore, we analyze the effects of time management strategies on the distribution of time over the moves of an average game, allowing us to partly explain their performance. In the experiments, the domain-independent strategy STOP provides a significant improvement over the state of the art in Go, and is the most effective time management strategy tested in all five domains.
|Journal||IEEE Transactions on Computational Intelligence and AI in Games|
|Publication status||Published - Sept 2016|
- Artificial intelligence
- game tree search
- Monte Carlo Tree Search
- time management