Improving Model and Search for Computer Go
This work aims to improve the performance of computer Go agents by exploring alternative network architectures and search algorithms, which is relevant for researchers and practitioners in game AI.
This paper explores the use of mobile networks as an alternative to residual networks in computer Go, demonstrating how network width and depth affect playing strength. It also introduces a generalized version of the PUCT search algorithm.
The standard for Deep Reinforcement Learning in games, following Alpha Zero, is to use residual networks and to increase the depth of the network to get better results. We propose to improve mobile networks as an alternative to residual networks and experimentally show the playing strength of the networks according to both their width and their depth. We also propose a generalization of the PUCT search algorithm that improves on PUCT.