Spikes that count: Rethinking spikiness in neurally embedded systems

Keren Saggie*, Alon Keinan, Eytan Ruppin

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

Abstract

Spiky neural networks are widely used in neural modeling, due to their biological relevance and high computational power. In this paper we investigate the usage of spiking dynamics in embedded artificial neural networks, that serve as a control mechanism for evolved autonomous agents performing a counting task. The synaptic weights and spiking dynamics are evolved using a genetic algorithm. We compare evolved spiky networks with evolved McCulloch-Pitts networks, while confronting new questions about the nature of "spikiness" and its contribution to the neurocontroller's processing. We show that in a memory-dependent task, network solutions that incorporate spiking dynamics can be less complex and easier to evolve than networks involving McCulloch-Pitts neurons. We identify and rigorously characterize two distinct properties of spiking dynamics in embedded agents: spikiness dynamic influence and spikiness functional contribution.

Original languageEnglish
Pages (from-to)303-311
Number of pages9
JournalNeurocomputing
Volume58-60
DOIs
StatePublished - Jun 2004

Funding

FundersFunder number
Tel-Aviv University
Israel Academy of Sciences and Humanities
Israel Science Foundation

    Keywords

    • Counting
    • Evolutionary computation
    • Neurocontroller analysis
    • Spiking

    Fingerprint

    Dive into the research topics of 'Spikes that count: Rethinking spikiness in neurally embedded systems'. Together they form a unique fingerprint.

    Cite this