Basic Analysis of Cellular Dynamic Binary Neural Networks

Abstract

This paper studies cellular dynamic binary neural networks that can generate various periodic orbits. The networks is characterized by signum activation function and local connection parameters. In order to visualize/analyze the dynamics, we present a feature plane of present two simple feature quantities. We also we present normal form equations that can describe all dynamics of the networks. Using the normal form equation and feature plane, various phenomena are investigated.