Non-ontogenic sparse neural networks
D. Elizondao, Emile Fiesler, Jerzy J. Korczak · 2002
Almost all artificial neural networks are by default fully connected, which often implies a large amount of redundancy and high complexity. Little research has been devoted to the study of sparse neural networks, with its potential advantages of reduced training and recall time, improved generalization capabilities, reduced hardware requirements, as well as being one step closer to biological reality. This publication presents a summary of the various kinds sparse neural networks, clustered into a lucid framework.