Reinforcement Learning through Active Inference

TL;DR

This paper introduces a novel reinforcement learning approach inspired by active inference, which inherently balances exploration and exploitation and offers a flexible view of reward, demonstrated through robust performance on various benchmarks.

Contribution

It develops a new decision-making objective called the free energy of the expected future, integrating active inference principles into RL.

Findings

Balances exploration and exploitation effectively

Performs well on benchmarks with sparse and no rewards

Provides a flexible conceptualization of reward

Abstract

The central tenet of reinforcement learning (RL) is that agents seek to maximize the sum of cumulative rewards. In contrast, active inference, an emerging framework within cognitive and computational neuroscience, proposes that agents act to maximize the evidence for a biased generative model. Here, we illustrate how ideas from active inference can augment traditional RL approaches by (i) furnishing an inherent balance of exploration and exploitation, and (ii) providing a more flexible conceptualization of reward. Inspired by active inference, we develop and implement a novel objective for decision making, which we term the free energy of the expected future. We demonstrate that the resulting algorithm successfully balances exploration and exploitation, simultaneously achieving robust performance on several challenging RL benchmarks with sparse, well-shaped, and no rewards.