Definition

artificial neuron

What is an artificial neuron?

An artificial neuron is a connection point in an artificial neural network. Artificial neural networks (ANNs), like the human body's biological neural network, have a layered architecture and each network node, or connection point, can process input and forward output to other nodes in the network.

In both artificial and biological architectures, the nodes are called neurons, and the connections are characterized by synaptic weights, which represent the significance of the connection. As new data is received and processed, the synaptic weights change, and this is how learning occurs.

Artificial neurons are modeled after the hierarchical arrangement of neurons in biological sensory systems. In the visual system, for example, light input passes through neurons in successive layers of the retina before being passed to neurons in the thalamus of the brain and then on to neurons in the brain's visual cortex. As the neurons pass signals through an increasing number of layers, the brain progressively extracts more information until it's confident it can identify what the person is seeing. In artificial intelligence (AI), this fine-tuning process is known as deep learning.

In both artificial and biological networks, when neurons process the input they receive, they decide whether the output should be passed on to the next layer as input. The decision of whether to send information on is called bias, and it's determined by an activation function built into the system. For example, an artificial neuron can only pass an output signal on to the next layer if its inputs, which are voltages, sum to a value above some particular threshold value. Because activation functions can either be linear or non-linear, neurons often have a wide range of convergence and divergence. Convergence is the ability for one neuron to receive input from many other neurons in the network, and divergence is the ability for one neuron to communicate with many other neurons in the network.

What is a neural network?

Broadly, a neural network is defined as a signal processing system consisting of interconnected nodes, through which input signals pass to generate an output. Each node receives signals from others preceding it in the network until it reaches an activation threshold, at which point it sends out a signal to other nodes downline. When the strength of the output of individual nodes is dynamically adjusted, the network can be trained to deliver specific outputs.

Neural networks are conceptually abstract, but they mimic the behavior of neurons, synapses and axons in the human brain. Biological neurons include brain cells and other nerve cells throughout the body that send signals to the brain. The idea behind the neural network concept is to process signals or information to generate desired outputs in the manner of the human brain.

The architecture of an artificial neural network.
A neural network is made up of input, output and hidden layers that are designed to mimic the processing power of the human brain.

What is an artificial neural network?

While biological neural networks are made up of biological neurons in human brains, ANNs are instantiated in software, running on digital computer systems -- essentially simulations of human brain activity, at least in principle. Artificial neurons, as described above, are the building blocks of such networks, arranged in layers that receive data and pass it forward to the next layer, once the activation threshold of each neuron is reached. A governing algorithm fine-tunes these activation thresholds across the network, training the system to improve its performance.

The granddaddy of these governing algorithms is the perceptron, a supervised learning mechanism originally designed for binary classification tasks. In its modern form, this algorithm is the foundation of machine learning systems, which in recent years have become the foundation of most AI applications.

Because the moving parts of an ANN are so easily reconfigured, many variations have been created over the past few decades. The following are the two most common:

  • Feedforward neural networks. This type of neural network is unidirectional. Input signals only flow in one direction, from the input side of the network to the output side. In such a network, signals don't cycle internally, and there are no looped processes. The governing algorithm in such a network is usually backpropagation. Errors in output trigger the fine-tuning of the activation thresholds of nodes in the network to optimize its output.
  • Convolutional neural networks. For more complex applications, this adaptation of the feedforward network architecture uses a cross-correlation technique in its weight adjustments to eliminate the backpropagation errors often seen in simpler networks.

What are some artificial neural network applications?

As mentioned above, ANNs are the platform of machine learning -- the foundation of modern AI. In recent years, the practical applications of machine learning have found many uses:

  • Image processing. Convolutional neural networks are often used for facial recognition by identifying individuals based on facial features and patterns in facial images. They can also be used to identify and classify objects including images or videos.
  • Pattern recognition. Feedforward neural networks are trained to associate input/output relations. This has applications in computer vision, speech recognition and handwriting.
  • Language translation. Deep learning and neural networks are often used to capture speech patterns that will be used to create algorithms that can translate text between different languages.
  • Healthcare. Artificial neural networks are used in medical diagnostics to identify diseases, including cancer. Algorithms are trained to detect cancerous tissue at the microscopic level.
  • Fraud detection. Large amounts of transaction data are analyzed to identify unusual patterns that could indicate fraudulent credit card activity.
  • Speech recognition. Personal assistants like Alexa, Siri and Cortana use natural language processing to interact with users.
  • Marketing and sales. E-commerce sites use AI to suggest other products and services to buy based on a person's viewing and previous buying histories.

Learn which AI and machine learning trends, including multimodal AI and shadow AI, are transforming the industry.

This was last updated in June 2024

Continue Reading About artificial neuron

Dig Deeper on Digital transformation

Cloud Computing
Mobile Computing
Data Center
Sustainability
and ESG
Close