West Saxon University of Applied Sciences Zwickau AI method yields valid results from incomplete data; less training data, more transparent results

West Saxon University of Applied Sciences Zwickau: New AI Method—Valid Results Despite Incomplete Measurement Data - Silicon Saxony

IBM: One in four data breaches is enabled by AI – Data breaches cost German companies an average of 4.25 million euros

Intel: Collaboration with Google Cloud to Accelerate Intel’s AI-Driven Transformation

SAP: Sovereign Data Infrastructure in Europe—Essential, or Not Even the Main Issue?

West Saxon University of Applied Sciences Zwickau: New AI Method—Valid Results Despite Incomplete Measurement Data

July 29, 2026. As part of his successful doctoral research at Friedrich Schiller University in Jena, Dr. Alexander Kabardiadi-Virkovski has developed a novel approach that specifically combines knowledge of physics with methods of artificial intelligence (AI). The research was conducted in the “Optical Technologies” research group at the West Saxon University of Applied Sciences in Zwickau (WHZ) and at the Fraunhofer Application Center for Optical Metrology and Surface Technologies (AZOM). The goal of the research is to reliably apply AI even in situations where measurement data is incomplete, noisy, or of limited usability. In the future, the developed methods could improve, among other things, industrial quality control, optical metrology, and medical applications.

Telefon: +49 351 8925 886

“Through his work, Alexander Kabardiadi-Virkovski demonstrates how physical knowledge and artificial intelligence can be meaningfully combined. This not only improves the transparency of AI models but also opens up new possibilities for industrial measurement technology,” says Prof. Dr. rer. nat. habil. Peter Hartmann, head of the Optical Technologies Research Group and of Fraunhofer AZOM.

AI Doesn’t Just Learn from Data

Many AI methods require large amounts of data and produce results whose origins are difficult to trace. Experts refer to these as “black-box” models.

This dissertation takes a different approach: Instead of training AI exclusively with measurement data, existing physical knowledge is specifically embedded into the architecture of the models. Physical relationships are not merely taken into account but are integrated into the model as structuring elements. As a result, the models require less training data while simultaneously delivering more transparent results. “I was interested in the question of how artificial intelligence and physics can be combined in such a way that reliable conclusions can be drawn even from incomplete measurement data. At the same time, the models’ decisions should remain transparent,” explains Dr. Alexander Kabardiadi-Virkovski.

The developed methodology was tested, among other things, in the analysis of light propagation in optical fibers, the evaluation of high-precision interferometric measurements, and the determination of material properties. The results show that physical expertise can be systematically integrated into modern machine learning methods, thereby opening up new application possibilities.

From WHZ Student to Research Group Leader

Dr. Alexander Kabardiadi-Virkovski’s academic career exemplifies the opportunities for professional development at WHZ. Since beginning his studies in 2009 in the Physical Engineering program, he has progressed from student to research assistant to a researcher with a Ph.D. and is now the head of a research group at Fraunhofer AZOM.

About the “Optical Technologies” Research Group at WHZ

The “Optical Technologies” research group at the Leupold Institute for Applied Natural Sciences at WHZ develops solutions for metrology tasks in fields such as medical technology, photonics, semiconductors, and automotive applications. The emphasis is on optical sensing, imaging, and data-driven analysis, with a clear focus on transferring these technologies into robust applications.

The Fraunhofer Application Center for Optical Metrology and Surface Technologies (AZOM) is a joint research facility of the West Saxon University of Applied Sciences Zwickau and the Fraunhofer Institute for Materials and Beam Technology (IWS). The center combines application-oriented university research with the expertise of the Fraunhofer Society and develops innovative solutions in the fields of optical metrology, image processing, artificial intelligence, and photonics.

Prof. Dr. rer. nat. habil. Peter Hartmann West Saxon University of Applied Sciences Zwickau Faculty of Physical Engineering

Computer Science Head of the Optical Technologies Research Group

Fraunhofer AZOM Peter.Hartmann[at]whz.de +49 375 536-1538 +49 1522 9262090

– – – – –