Key-Threshold based spiking neural network

Abstract

In the paper a novel model of Key-Threshold based Spiking Neural Network (KTSNN) is proposed. This neural network consists of quasi-neurons oriented to recognize any key-spikes distributed in time (sequence of spikes) or in space (in synapses). Every neuron aims to recognize key (template of spikes) stored in its memory and to react by output spike at successfully detection. Software implementation of this model is suggested. Possible methods of learning, implementation and usage of this model are discussed.