How programmable is the artificial neural network

How do artificial neural networks work?

The human brain is a model for artificial neural networks. These networks can learn to act independently and thus help to implement automated driving. But how do they work?


Artificial neural networks (ANN) simulate information processing based on the model of the human brain. The reason for this is that "conventional computer programs are not well suited for all tasks, while human thinking can solve them without any problems", explain the Springer authors Styczynski, Rudion and Naumann in the chapter Artificial Neural Networks from the book Introduction to Expert Systems. Examples of this are pattern recognition, speech recognition or prognoses.

Accordingly, ANNs are developed on the model of the human brain and trained using a large amount of data. "The aim of artificial neural networks is to simulate the principle of learning in the form of a computer program," summarizes Springer author Sebastian Dörn in the chapter on neural networks from the book Programming for Engineers and Scientists.

01.12.2019 | Development | Edition 12/2019

Machine learning for automated driving

Automated and autonomous driving in SAE Levels 3 to 5 are among the core topics of future mobility. In its "Park and Charge" innovation project, Bertrandt shows how important it is to recognize surroundings and precisely plan trajectories with the help of artificial intelligence. Machine learning is used to improve localization, networking and cloud applications.

The learning then works as follows: New skills are learned through imitation using the principle of trial and error. As in the human brain, the neuron and the networking of neurons also play a decisive role in ANNs. The artificial neurons, also called nodes, are arranged in layers that fulfill different functions. "Each layer consists of a large number of individual neurons, the number of which varies depending on the architecture. Each neuron in a layer has an architecture-based network with the neurons of the next layer," explains Bertrandt in the article Machine Learning for Automated Driving from ATZ 12-2019. A fully connected layer (FCL) is used when every neuron in one layer is connected to all neurons in the next layer. The intermediate layers are referred to as hidden layers, as typically only the first and last layers are addressed for input and output.

The more data, the better

When learning, the networks change all information or numerical values ​​at the connections between the nodes until the results are good enough. "In their inner layers, the networks independently develop compact representations from the raw data, which means that many preprocessing programs are superfluous and the actual task is easier to learn," explains Springer author Verena Fink in the chapter Monitoring or reinforcing - learning methods in comparison from the book Quick Guide KI -Projects - just do it. The model developed by the KKN can then be applied to new, potentially unknown data of the same type.

In general, the following applies: The number of layers determines, among other things, the degree of complexity that an artificial neural network can map. KKN, which have a particularly large number of hidden layers between the input and output layers, make a neural network "deep", which leads to the concept of deep learning. And the more data that is available, the better the neural network will work. "Big data and artificial intelligence complement each other perfectly here: Neural networks need big data in order to be well trained, and big data require new computer models (especially neural networks), since conventional computers can no longer process the large amounts of data efficiently enough", so the Springer authors Kohn and Tamm in the chapter Introduction to Neural Networks from Mathematics for Business Information Systems.


AI helps with automated driving

Machine learning and neural networks play a major role in software development and automated driving. "Examples of the use of AI solutions are the derivation of driving strategies from large amounts of input data from unstructured environments or the prediction of traffic situations with a time horizon of several seconds," says Elektrobit in the article How neural networks change the development of automotive software from ATZelectronics 1-2-2020 on.