Implementation of inhibitory operation in neuromorphic system

Abstract

An inhibition part of synapse-neuron connection is indispensable in a neuromorphic system for hardware implementation of artificial intelligence. First, when the system operates with the winner takes all method, a lateral inhibition is required to permit one neuron firing, thereby preventing other neurons from firing. In addition, in order to implement the negative weight, an inhibition part, which is the role of subtracting the signal, must be accompanied. Therefore, in this paper, the structure of a simple inhibition part for lateral inhibition and negative weight is presented and it is verified by simulation.