Memory Control in a FitzHugh-Nagumo Network via STDP

Abstract

The present paper introduces a network architecture based on FitzHugh-Nagumo neurones with connection weights adjusted via Spike Time Dependent Plasticity (STDP) and discrete transmission delays. Networks are trained to encode spatial patterns of spiking cells. It is shown that multiple patterns can be encoded within the network by this method and that pattern recall can be achieved by stimulating a subset of the neurones. Furthermore, we show that increasing the number of cells participating in a pattern can favourably improve recall performance and reliability. Overall, these results show that spiking neural networks with transmission delays and STDP enable the control of network behaviour for the encoding of spatial associative memories.


Back to Table of Contents