A Spiking Neural Learning Classifier System

TL;DR

This paper introduces a novel learning classifier system using spiking neural networks with dynamic structures, capable of solving complex continuous input problems and sequencing actions effectively, demonstrated on a robotics platform.

Contribution

It presents a new LCS where rules are spiking neural networks with constructivist growth, enabling flexible, temporal, and hierarchical learning.

Findings

Successfully solved a continuous input problem with complex structure.

Enabled temporal state decomposition for sequence learning.

Performed effectively on a simulated robotics platform.

Abstract

Learning Classifier Systems (LCS) are population-based reinforcement learners used in a wide variety of applications. This paper presents a LCS where each traditional rule is represented by a spiking neural network, a type of network with dynamic internal state. We employ a constructivist model of growth of both neurons and dendrites that realise flexible learning by evolving structures of sufficient complexity to solve a well-known problem involving continuous, real-valued inputs. Additionally, we extend the system to enable temporal state decomposition. By allowing our LCS to chain together sequences of heterogeneous actions into macro-actions, it is shown to perform optimally in a problem where traditional methods can fail to find a solution in a reasonable amount of time. Our final system is tested on a simulated robotics platform.