Transfer of Fully Convolutional Policy-Value Networks Between Games and Game Variants

TL;DR

This paper demonstrates how fully convolutional policy-value networks can be transferred across different game variants and games using shared semantic representations, enabling zero-shot transfer and fine-tuning.

Contribution

It introduces a method for transferring trained convolutional networks between game variants and different games based on shared semantics of state and action representations.

Findings

Successful zero-shot transfer between game variants

Effective fine-tuning improves performance on new game variants

Transfer learning reduces training time for new game variants

Abstract

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.