A spiking neural learning classifier system

Howard, Gerard David; Bull, Larry; Lanzi, P. L.

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.

• Journal Article Type: Article
• Journal: arXiv preprint arXiv:1201.3249
• Peer Reviewed: Peer Reviewed
• Keywords: robotics, learning, computer science, neural and evolutionary computing
• Public URL: https://uwe-repository.worktribe.com/output/950424
• Publisher URL: http://arxiv.org