A model of spike neuron oriented to hardware implementation

Abstract

One of most perspective and popular area of neural networks is neuromorphic computing dealing with development of brain-like spike neural networks oriented on hardware implementation. Such neural networks will provide replacement of computers with Von Neumann architecture in such fields as computer vision and control of autonomous robots. In this paper we suggest one model of spike integrate and fire neuron for development of such neural networks. This model is distinguished by simple arithmetic operations, providing, short-long controlled memory of integrated input signals and controlled refractory period for output. Proposed model will provide simple hardware implementation.