Skip to main content

Artificial Neural Network (ANN)

What is an Artificial Neural Network?

An artificial neuron network (ANN) is a computing system modeled after the operation of neurons in the human brain.

How Do Artificial Neural Networks Work?

Artificial Neural Networks (ANNs) are mathematical models capable of learning, making them a key component in artificial intelligence (AI), machine learning (ML), and deep learning. ANNs are best viewed as weighted directed graphs, organized in layers of interconnected nodes that imitate biological neurons of the human brain. The first layer receives raw input signals, similar to optic nerves in human visual processing. Each successive layer receives output from the preceding layer, similar to the way neurons that are further from the optic nerve receive signals from those closest to them. The output at each node is called its activation or node value. The last tier produces the system's output.

Perceptron Artificial Neural Network

The Perceptron is the simplest type of ANN, typically used for making binary predictions. A Perceptron can only work if the data can be linearly separable.

Multi-layer Artificial Neural Network

A fully connected multi-layer neural network, also known as a Multilayer Perceptron (MLP), is made of more than one layer of artificial neurons or nodes. This type of ANN is used to solve more complex classification and regression tasks. The most common model is the 3-layer fully-connected backpropagation model. The first layer consists of input neurons that send data to the second layer, which sends output neurons to the third layer.

Furthermore, there are two Artificial Neural Network topologies: FeedForward and Feedback.

FeedForward Artificial Neural Network

In this ANN, the information flow is unidirectional. The information travels only in one direction; forward; without making any feedback loops. It first goes through the input nodes, then through the hidden nodes (in case there are any), and in the end, it goes through the output nodes.

Feedback Artificial Neural Network

In this case, there are inherent feedback connections between the neurons of the networks. Feedback loops are allowed, making this topology useful for tasks such as speech recognition and image processing.