Improving Performance of Spike Pattern Detection Using Close-to-Biology Spiking Neuronal Network

Takuya Nanami, Takashi Kohno

Abstract: In the nervous system, there is a broad variety of neuron types, each exhibiting distinct firing properties. Although these neurons are considered important, the understanding of their role in information processing remains limited. In this study, we constructed a simple network using a piecewise quadratic neuron (PQN) model that can reproduce a variety of neuronal activities. Further, we examined the effect of various neuronal dynamics on the success rate of a biologically plausible spike-pattern detection task. The simulation results showed that certain mathematical structures increased the success rate of spike-pattern detection.