Deep Learning Dictionary - Lightweight Crash Course

Neural Network Layers - Deep Learning Dictionary

An artificial neural network is made up of multiple processing units called nodes that are organized into layers. These layers are connected to each other via weights.

A general network consists of an input layer, which receives the input data, an output layer, which supplies the network predictions on the given input, and hidden layers, which process the data during a forward pass.

There are many types of layers that process data in different ways and are chosen and used for specific purposes.

For many computer vision tasks in deep learning, we make use of:

• Fully connected layers (Also known as dense and linear layers)
• Convolutional layers

When data is passed as input to a given layer, the nodes in that layer process the input and send the output to the following layer.

Different layers perform different transformations on their inputs, and some layers are better suited for some tasks than others. For example, convolutional layers are the most popular layer type used in networks that work with image data.

A fully connected layer is one that is fully connects each of its inputs to each of its outputs via weights. This can be shown as each node in a given layer being fully connected by weights to all nodes in the previous layer and all nodes in the following layer.

There are various types of networks, and actually, the network type is often categorized by its layer types and how these layers are organized within the network. For example, a convolutional neural network (CNN) is a network with convolutional layers.