Self-organization of action hierarchy and compositionality by reinforcement learning with recurrent neural networks

TL;DR

This paper introduces a novel stochastic RNN architecture for reinforcement learning that autonomously develops action hierarchies and compositionality, leading to improved learning efficiency and re-learning in complex tasks.

Contribution

The paper presents a new multiple-timescale stochastic RNN that self-organizes action hierarchies and compositionality, advancing understanding of neural mechanisms in RL.

Findings

The network can learn to abstract sub-goals autonomously.

Self-developed compositionality accelerates re-learning on new tasks.

Stochastic neural activities improve overall performance.

Abstract

Recurrent neural networks (RNNs) for reinforcement learning (RL) have shown distinct advantages, e.g., solving memory-dependent tasks and meta-learning. However, little effort has been spent on improving RNN architectures and on understanding the underlying neural mechanisms for performance gain. In this paper, we propose a novel, multiple-timescale, stochastic RNN for RL. Empirical results show that the network can autonomously learn to abstract sub-goals and can self-develop an action hierarchy using internal dynamics in a challenging continuous control task. Furthermore, we show that the self-developed compositionality of the network enhances faster re-learning when adapting to a new task that is a re-composition of previously learned sub-goals, than when starting from scratch. We also found that improved performance can be achieved when neural activities are subject to stochastic rather than deterministic dynamics.