The CNN: The Architecture Behind Artificial Intelligence Development

Abstract

Listen

Translate article

The convolutional neural network (CNN) is a multilayer network architecture that is capable of training itself using an advanced algorithm to produce increasingly accurate results. The CNN is especially effective in spatial image recognition and is used in a multitude of fields, such as image recognition in the medical industry, and image segmentation in the security field. The ability of the CNN to show impressive results is seen in its multilayer composition.
 This multilayer network consists of the convolutional layer, the pooling layer, the activation layer, and the fully connected layer. Here, the convolutional layer, the pooling layer, and the activation layer have their own parameters which will influence the flow of input data throughout the CNN until it produces an ultimate output in the connected layer. As such, this paper will delve into the individual characteristics of each layer, and introduce its relationship with not only its immediate surrounding layers but with the entirety of the CNN. Additionally, this paper will stress important modules and parameters that can help improve the layers.