Color image classification on neuromorphic system using reinforcement learning

Abstract

Neuromorphic hardware system has been studied based on binary monochromatic numerical data. In this study, the input to neuromorphic hardware systems uses binary color images rather than monochrome. Color image classification is implemented using reinforcement learning. In addition, the smaller degree of freedom of synaptic weights is experimented to implement a binary neuromorphic hardware system. More complicated color image classification can be developed with this study.