Ilmenau researchers are developing the safe use of AI in critical systems

Ilmenau researchers are developing the safe use of AI in critical systems

The Technical University of Ilmenau has integrated itself into a pioneering research group of the German Research Foundation (DFG) that deals with the challenges of machine learning. Their goal is to increase the security, performance and data efficiency of this technology. The focus is particularly on complex control systems and safety-critical applications that cover essential areas such as electrical energy systems, autonomous driving and robotics. Ilmenau mathematics professor Karl Worthmann will lead the interface between machine learning and mathematical precision.

The research project, which is coordinated by Leibniz University Hannover, started at the end of April with an international workshop. Almost 3.5 million euros in funding will be made available for the next four years, of which 486,400 euros will go directly to the TU Ilmenau. This initiative highlights the central role of machine learning in the development of artificial intelligence and the limits that traditional systems reach in terms of security requirements during ongoing operation.

Integration into industry

Another crucial aspect of current developments is the integration of machine learning into industrial applications. A newly established in-house training course entitled “Machine Learning for safety-critical applications in industry” aims to take specific safety principles into account. This is done taking into account existing and future standards for artificial intelligence as well as industry-specific standards.

Training participants learn about the effects of machine learning on functional safety. This also includes the application of key concepts such as robustness, bias and prediction certainty as well as the development of a project-specific safety life cycle. Engineers who need to integrate AI into production processes and provide safety evidence are among the target audience for this training. The program provides a structured framework and a comprehensive toolkit for the secure use of machine learning.

Technological basics

The fundamental element that supports machine learning is neural networks, a subset of this technology. Neural networks are heavily inspired by the connections between nerve cells in the human brain. These artificial networks consist of multiple layers of data nodes linked together by weighted connections. By repeatedly presenting data, the neural network learns to classify the information more effectively.

The adjustment of the weights between the neurons occurs continuously, making the created model applicable to unknown data. A special type of neural networks are the so-called “deep neural networks”. These networks can contain hundreds of thousands or even millions of layers of neurons, enabling increasingly complex problems to be solved through deep learning. Continuous training and adaptation of connections are crucial to the success of these learning processes.

In summary, it can be said that the Technical University of Ilmenau and its partners make a significant contribution to the further development of machine learning and its safe application in critical areas through innovative approaches and training programs.

For further information about the research group at TU Ilmenau, visit tu-ilmenau.de. You can find detailed insights into in-house training at iks.fraunhofer.de and additional information on artificial intelligence and neural networks at iks.fraunhofer.de.