Accelerating Empowerment Computation with UCT Tree Search

TL;DR

This paper introduces a modified UCT tree search method to efficiently compute empowerment, an intrinsic motivation metric, enabling believable agent behaviour in video game scenarios without heavy computational costs.

Contribution

It proposes a novel UCT-based algorithm with modifications to approximate empowerment maximisation efficiently in deterministic environments.

Findings

Approximates empowerment close to exhaustive computation with fewer resources.

Enhances sampling efficiency through three key modifications.

Produces believable, intrinsic motivation-driven behaviour in a Minecraft-like environment.

Abstract

Models of intrinsic motivation present an important means to produce sensible behaviour in the absence of extrinsic rewards. Applications in video games are varied, and range from intrinsically motivated general game-playing agents to non-player characters such as companions and enemies. The information-theoretic quantity of Empowerment is a particularly promising candidate motivation to produce believable, generic and robust behaviour. However, while it can be used in the absence of external reward functions that would need to be crafted and learned, empowerment is computationally expensive. In this paper, we propose a modified UCT tree search method to mitigate empowerment's computational complexity in discrete and deterministic scenarios. We demonstrate how to modify a Monte-Carlo Search Tree with UCT to realise empowerment maximisation, and discuss three additional modifications that facilitate better sampling. We evaluate the approach both quantitatively, by analysing how close our approach gets to the baseline of exhaustive empowerment computation with varying amounts of computational resources, and qualitatively, by analysing the resulting behaviour in a Minecraft-like scenario.