Introduction to Convolutional Neural Networks (CNNs) | The Most Popular Deep Learning architecture

Introduction

Yann LeCunn introduced the concept of Convolutional Neural Networks (or CNNs) in 1998 with its paper “Gradient-Based Learning Applied to Document Recognition”. It was later popularized in 2012 by Alex Krizhevsky winning the ImageNet competition wherein he used the networks to drop the image classification error from 26% using computer vision techniques to 15% with CNNs. This was a substantial drop and was considered a turning point in the history of digital image classification. It was popularized much later because of the insufficient power of the GPUs at the time. It is the exponential growth over the years of these GPUs made the use of deeper networks like CNNs possible 14 years later and to this day.

A General Description