Convolutional Neural Networks: I

Hubel and Wiesel proposed that the pattern recognition tasks in monkeys and cats use two types of cells, one of which has a larger receptive field. The output of this field does not depend on the location of the edges in the field. This inspired Kunihiko Fukushima to introduce neo-cognition, which in turn inspired convolutional and downsampling layers in Neural Networks, called Convolutional Neural Networks (CNNs). Backpropagation was used in the CNNs by Yann LeCun, a French computer scientist and the recipient of the prestigious Turing Award. LeNet, the first CNN, could recognize handwritten digits. Ignored initially, the CNNs got their due share in 2012, with the advent of AlexNet. CNNs have been successfully applied to image classification, object detection, and disease prediction and even in digital arts. They have shown better performance compared with the existing Neural Networks and are being extensively used in numerous disciplines.