Review of the Convolution Neural Network Architectures for Deep Learning

Abstract

Convolution Neural Network (CNN) architectures have been significantly evolved during the last two decades since 1998.The field of Deep learning (DL)is significantly improved over time where the new ideas implemented to have bettercomputing power performance for divergedapplications.In this paper, the fundamental math flow behind CNN based classification is presented and the CNN architectures evolved so far are reviewed. Moreover, the improvements are presented in chronological order. This article helps the beginner to have the over view of CNNs and architectures from the scratch.
Keywords: Convolution Neural Network (CNN); Deep learning (DL); ILSVRC; CNN Architecture