Transfer of Fully Convolutional Policy-Value Networks Between Games and Game Variants
This work addresses transfer learning challenges in game AI, but it is incremental as it builds on existing AlphaZero and Ludii frameworks.
The paper tackled the problem of transferring trained parameters between board games and their variants using fully convolutional architectures in AlphaZero-like self-play setups, achieving results through zero-shot transfer and fine-tuning experiments with Ludii's game library.
In this paper, we use fully convolutional architectures in AlphaZero-like self-play training setups to facilitate transfer between variants of board games as well as distinct games. We explore how to transfer trained parameters of these architectures based on shared semantics of channels in the state and action representations of the Ludii general game system. We use Ludii's large library of games and game variants for extensive transfer learning evaluations, in zero-shot transfer experiments as well as experiments with additional fine-tuning time.