Convolution Neural Network-Based Architecture

Welcome back readers !!! Today's article is a brief description of convolution neural network.

Convolution neural network (CNN) is inspired by the multi-layer perceptrons (MLP) model. MLP model is made of multiple layers of artificial neurons with weights associated with them individually. In CNN, these weights are shared by neurons to reduce feature dimensionality that helps the model to converge faster. This network is implemented in computer vision-based problems such as transfer style, image regression, and image classification. It consists of convolution layer, pooling layer, and fully connected layers.

As said in the previous article, there are three types of CNN namely, traditional convolution-based networks, auto-encoders, and end-to-end networks. These will be discussed after a few articles, i.e., after the understanding of deep learning layers.

Akshay Juneja authored 15 articles for INFO4EEE Website on Deep Learning.