Abalearn: A risk-sensitive approach to self-play learning in Abalone

Abstract

This paper presents Abalearn, a self-teaching Abalone program capable of automatically reaching an intermediate level of play without needing expert-labeled training examples, deep searches or exposure to competent play. Our approach is based on a reinforcement learning algorithm that is risk-seeking, since defensive players in Abalone tend to never end a game. We show that it is the risk-sensitivity that allows a successful self-play training. We also propose a set of features that seem relevant for achieving a good level of play. We evaluate our approach using a fixed heuristic opponent as a benchmark, pitting our agents against human players online and comparing samples of our agents at different times of training.