The N-Tuple Bandit Evolutionary Algorithm for Game Agent Optimisation

TL;DR

The paper introduces NTBEA, an efficient optimization algorithm for noisy, costly discrete problems, demonstrating superior performance in game hyper-parameter tuning compared to grid search and EDA.

Contribution

It presents the NTBEA algorithm, a novel, simple, and effective method for optimizing noisy discrete problems, especially in game hyper-parameter tuning.

Findings

NTBEA outperforms grid search in noisy settings.

NTBEA surpasses estimation of distribution algorithms.

The model efficiently approximates fitness and evaluation counts.

Abstract

This paper describes the N-Tuple Bandit Evolutionary Algorithm (NTBEA), an optimisation algorithm developed for noisy and expensive discrete (combinatorial) optimisation problems. The algorithm is applied to two game-based hyper-parameter optimisation problems. The N-Tuple system directly models the statistics, approximating the fitness and number of evaluations of each modelled combination of parameters. The model is simple, efficient and informative. Results show that the NTBEA significantly outperforms grid search and an estimation of distribution algorithm.