Artificial neural network: what are their capacities?
A big computer dream could soon see the light of day: computers with capacities far superior to those of the human brain. In recent years, research on artificial intelligence has made enormous progress. Artificial neural networks are a crucial technology for enabling machines to learn and "think" independently.
Artificial neural networks allow computers to solve problems on their own and enhance their abilities in general. Some require initial supervision, depending on the artificial intelligence method used.
The design of artificial neural networks is based on the structure of biological neurons in the human brain.
Artificial neural networks can be described as systems made up of at least two layers of neurons - an input layer and an output layer - and typically including "hidden layers". The more complex the problem to be solved, the more layers the artificial neural network must have. Each layer contains a large number of specialized artificial neurons.
Processing of information within the neural network
Within an artificial neural network, information processing always follows the same sequence: information is transmitted as signals to neurons at the input layer, where it is processed. Each neuron is assigned a particular "weight", and therefore a different importance. Combined with the so-called transfer function, the weight determines what information can enter the system.
In the next step, a so-called activation function associated with a threshold value calculate and weight the output value of the neuron. Depending on this value, a greater or lesser number of neurons are connected and activated.
This connection and this weighting draw an algorithm that matches a result to each entry. Each new iteration adjusts the weighting and therefore the algorithm so that the network gives a more precise and reliable result each time.
Artificial neural network: an example of an application
Artificial neural networks can be used for image recognition. Unlike the human brain, a computer cannot determine at a glance whether a photograph shows a human being, a plant or an object. He is forced to examine the image to discern its individual characteristics. It is the algorithm put in place that allows him to know which characteristics are relevant; otherwise, he can find out for himself through data analysis.
Within each layer of the neural network, the system checks the input signals, that is, the images, broken down into individual criteria such as colour, angles or shapes. Each new test allows the computer to more accurately determine what the image shows.
Initially, the results are necessarily subject to a number of errors. If the neural network receives human-made feedback that allows it to adapt its algorithm, it is called machine learning. The concept of deep learning aims to eliminate the need for human intervention. The system then learns from its own experience; it improves each time an image is submitted to it.
The different types of artificial neural networks
Different structures of artificial neural networks are used depending on the learning method used and the desired objective.
Originally, the simplest form of artificial neural network consisted of a single neuron modified by weights and with a threshold value. The term "Perceptron" now also refers to single-layer forward propagation networks.
Forward propagation neural networks
A forward propagating artificial neural network can only convey information in one direction of processing. Networks can be single-layered, that is, consisting of only input and output layers, or multi-layered, that is, having a number of hidden layers.
In theory, we obtain on arrival an algorithm capable of identifying without error the content of a photograph, whether in colour or in black and white, whatever the position of the subject or the angle at which it is viewed. is represented.
Recurrent neural networks
In recurrent neural networks, it is possible to pass information through feedback loops, and thus return it to a previous layer. These feedbacks allow the system to build up a memory. Recurrent neural networks are used, for example, in speech recognition, translation and handwriting recognition.
Convolutional neural networks
A convolutional neural network is a type of multi-layered network. It is made up of a minimum of five layers. Pattern recognition is carried out on each of these layers. The result obtained on each layer is transmitted to the next layer. This type of artificial neural network is used in image recognition.
To ensure that the connections within an artificial neural network are properly established, it is first necessary to "train" it. Two basic procedures can be distinguished here: supervised learning and unsupervised learning.
As part of supervised learning, a concrete result is defined for each of the different entry options. For example, if the system is presented with a photograph of cats for recognition, it is necessary to monitor the operation of the system and provide feedback to determine whether the image is correctly recognized or not. This method makes it possible to modify the weights within the artificial neural network and to optimize the algorithm.
In the case of unsupervised learning, the result of the task in question is not predetermined. It is the system itself that draws consequences from the information entered alone, relying on Hebb's learning rule or adaptive resonance theory.
Artificial neural networks and their applications
Artificial neural networks are a powerful tool in cases where you are faced with a large amount of data without first knowing where the solution should be headed. They are typically used in handwriting, image, and voice recognition, where a computer system looks for certain characteristics in order to assign them.
It is also possible to use artificial neural networks to carry out all types of predictions or simulations. This is the case, for example, with weather forecasts, medical diagnoses or stock markets.
In industry, artificial neural networks are sometimes used as part of activity monitoring technologies, to detect any deviations from determined values and automatically take the necessary countermeasures, or to independently set target values taking into account the data evaluation carried out by the networks.
The development of unsupervised learning of artificial neural networks can now significantly expand their performance and scope. Among the most important applications of machine learning artificial neural networks are the well-known examples of Alexa, Siri and Google's voice assistant.