Deliverables

MCTS is a widely used search algorithm for developing agents that can play board games intelligently. Over the past two decades, researchers have proposed dozens, if not hundreds, of MCTS variants. Despite this, it's been challenging to determine which variants are best suited for specific types of games. In most studies, researchers demonstrate that a new MCTS variant outperforms one or a few other variants in a limited set of games. However, it’s uncommon for a new variant to consistently outperform others across a broad range of games, making it unclear which types of games certain MCTS variants are best at. Answering this question would greatly improve our understanding of MCTS algorithms, and help us make better decisions about which variants to apply to new games or other decision-making problems. This competition challenges you to develop a model that can predict the performance of one MCTS variant against another in a given game, based on the features of the game. Your work could help pave the way for identifying the strengths and weaknesses of different MCTS variants, advancing our understanding of where they work best in various scenarios.

Month 10

The Action will maintain a public website to coordinate and bundle all dissemination activities at this address: gametable.network. All events planned by the Action will be announced there. Job opportunities, teaching materials, and publications will be available for download.

Month 2