Explainer: artificial neural networks

AI Artificial intelligence neural network deep machine learning concept. Young business woman pressing button on virtual screen.
Posted by

Jan 12, 2024

From every Google search we make to the Alexa and Siri apps on our smartphones and tablets to the doctor’s office, artificial neural networks are increasingly transforming the way we live. But what exactly are they, and how do they work?

What is an artificial neural network?

Put simply, an Artificial Neural Network (ANN) is an efficient computing system that attempts to understand things and make decisions in a human-like way. An ANN is a form of artificial intelligence (AI), and more specifically, a type of machine learning. It is so named because it is loosely based on the concept of how neurons (brain cells) work inside a living brain and nervous system.

ANNs are a large collection of units (also known as nodes, or neurons) interconnected in a pattern that allows the units to talk to each other. Each artificial neuron receives signals then processes them and can signal to neurons connected to it.

Every connection link (or ‘edge’) between neurons is associated with a ‘weight’ containing information about the input signal. As data are fed into it and the ANN works on solving a problem, the weight usually either increases or decreases the signal that is being communicated. Neurons also have an internal state, which is called an activation signal. When input signals and the activation rule are combined and a specified threshold value within the node is reached, an output signal is produced that can then be sent to other neurons. If the threshold value isn’t reached, no signal is passed along to other neurons.

Typically, neurons are aggregated into layers. Different layers may perform different transformations on their inputs. Signals travel from the first layer (the input layer), through one or more hidden layers to the last layer (the output layer), sometimes after going back and forth through the layers many times.

As our human brains learn, the connections between neurons are either formed, changed or removed – and in a similar way, an ANN adjusts its weights to account for each new set of data used to ‘train’ it. As with the way people learn, practice makes perfect, so each new data set helps the ANN to learn and become better at what it is doing. Once these so-called learning algorithms are fine tuned for accuracy, they are powerful tools in computer science, allowing for high speed classifying and clustering of data.

ANNs are at the heart of a type of machine learning called deep learning. By manipulating data through a variety of neural networks, people can build powerful tools that can do some incredible things.

What are some examples of ANNs and what are they used for?

 “If you used Google to find this article, you used Google’s neural network that ranks the most relevant pages based on the keyword(s) you gave it. If you recently went on Amazon.com, all the recommended products that the website suggested to you was curated by a neural network. Even today, if you used your phone, you probably encountered a neural network that made your life easier. It’s all around us, and they all do different things and work in different ways,” says Towards Data Science’s Vansh Sethi.

Social media uses ANN data analysis heavily – for example, in Facebook’s ‘people you may know’ feature which suggests people you might know in real life so you can connect with them. Behind the scenes, ANNs are analysing data from your profile on your interests, your friends and their friends to work out the people you may know.

Facial recognition software found on photo apps or on social media is another example of machine learning using ANNs, which works by finding around 100 reference points on the human face and matching them with reference points already held in its database to identify people in an uncannily accurate way.

Healthcare – ANNs are a burgeoning and exciting field in medicine with the potential to enhance the diagnostic abilities of medical experts, save time and resources and ultimately improve the quality of medical care worldwide. As an example, ANNs have been trained to spot patterns in standard scans of the back of the eye, processing thousands of individual patient scan images to diagnose the early stages of the eye disease geographic atrophy. Where manual diagnosis by an expert would take hours, ANNs can do the job in minutes or seconds. Neural networks have also been used in cancer diagnosis to train algorithms that can identify cancerous tissue at the microscopic level at the same level of accuracy as specialist doctors. Some rare diseases can also be diagnosed by ANN facial analysis on patient photos. 

Siri and Alexa – the digital personal assistants whose dulcet tones can be heard on your phone or tablet can recognise speech using Natural Language Processing to interact with users and come up with a response. Natural Language Processing uses ANNs made to handle many tasks of these personal assistants such as managing the language syntax, semantics and correct speech.

What are the different types of artificial neural network?

While there are many varieties of ANN, some of the most commonly used types are:

The perceptron – the oldest and simplest form of a neural network, with a single artificial neuron. Perceptrons were first introduced by American psychologist Frank Rosenblatt in 1957 at Cornell Aeronautical Laboratory. Inspired by the biological neurons in our brains, Rosenblatt’s version consisted of one or more inputs, a processor and only one output.

Feedforward neural networks, or multi-layer perceptrons (MLPs) – these are the type of ANN described in the example above, with an input layer, a hidden layer or layers, and an output layer. Data usually is fed into these models to train them, and they are the foundation for computer vision, natural language processing, and other neural networks.

Convolutional neural networks (CNNs)   usually used for image recognition, pattern recognition and other types of computer vision. They harness principles from linear algebra, particularly matrix multiplication, to identify patterns within an image.

Recurrent neural networks (RNNs) – these are identified by their feedback loops. These learning algorithms are generally applied when using time series data to make predictions about future outcomes, such as sales forecasting or stock market predictions. 

Learn specialist skills in a fast growth area of computer science

The use of Artificial Intelligence (AI) in the modern business landscape is a fast growth area as organisations continue to recognise the impact it can have on their future success. There’s never been a better time to specialise in this exciting area.

The University of Abertay’s MSc Computer Science with Artificial Intelligence will equip you with a firm foundation in the field of computer science alongside specialised knowledge of the rapidly evolving field of AI. The MSc builds on our staff expertise in the area and has been developed based on Abertay’s strong connections with industry, so the cutting-edge skills and knowledge you will acquire will prepare you for a successful future.

You can start your studies within weeks, as there are six starts a year. What is more, the MSc is taught entirely online and part-time, so you can continue to build your career as you fit your studies around it.