In the course of 2023, the Institute of Computer Science will hire a total of four junior research group leaders, which will expand the range of research opportunities at Paderborn University for students as well.
Since the winter semester 2022/23, the Institute of Computer Science can welcome a new junior research group leader with Dr.-Ing. Philipp Terhörst. With the research field "Responsible Artificial Intelligence for Biometrics", students can look forward to an exciting topic that has so far only been dealt with to a limited extent at the university. Biometric systems are designed to verify a person's identity and are used, for example, in smart homes for voice recognition or for facial recognition at airports.
Dr.-Ing. Philipp Terhörst first studied physics, with a minor in computer science, at the TU Darmstadt. Already at that time, he was fascinated by the mixture of physics, computer science, and mathematics. A former tutor finally introduced him to the topic of machine learning, which resulted in a master's thesis at the Fraunhofer Institute for Computer Graphics in Darmstadt. This is also how he came to his current research focus on biometrics, where biological data is measured and analyzed. He then focused on the topic of face recognition, which is still part of his research today. He received his PhD on the topic of "Mitigating Soft-Biometrics Driven Privacy and Bias Concerns in Face Recognition Systems." His work won him the IANUS prize, the European Biometrics Industry Award, given by the European Association for Biometrics, and his dissertation was named the best at TU Darmstadt. Subsequently, he received the Alain Bensoussan Fellowship of the European Research Consortiums for Informatics and Mathematics, resulting in a stay abroad at the Norwegian University of Science and Technology.
As the new head of the junior research group, Dr.-Ing. Philipp Terhörst focuses on the four main topics of fairness, privacy, reliability and explainability. The research focus "Fairness" is about ensuring that all groups of people are treated equally by the algorithm and that there is no discrimination due to larger error rates. One problem is that a lot of information is processed that is not necessary for the application. Therefore, he is already developing fair algorithms that are not affected by this problem. Since other information can be derived in addition to the identity, but which is not needed for the recognition purpose, this also gives rise to data protection problems, to which he would like to devote more attention with the topic of "Privacy". Dr.-Ing. Terhörst emphasizes that biometric systems might have a strong impact on users, which can be associated with financial and social consequences. The "Reliability" topic therefore focuses on the reliability of the algorithms. It is also important to him to make the biometric recognition systems more explainable so that humans can understand the decision-making process to trust the system's decision or not. He discusses this under the topic "Explainability".
Another goal is to network the scientific community working on the topic of biometrics more closely and to draw attention to the topic of responsibility and to Paderborn as a location. Setting up a research group also gives him the opportunity to drive research forward much faster and in greater depth. In the coming years, he wants to build biometric facial recognition systems that are not only tailored to industry, but also address societal needs. The fact that his developments can be directly incorporated into practice is motivating and nice to see for Dr.-Ing. Terhörst: "I try to solve problems that occur in practice and have direct consequences on reality.". In addition to his own research, Dr.-Ing. Terhörst also looks forward to teaching: "What I really like is working with students. My experience is that they are incredibly motivated, unbiased, and open-minded.". Currently, he is teaching the master's course "Machine Learning for Biometrics.". This will be followed in the summer semester by a bachelor course, "Introduction to Human-Centered Machine Learning".
The junior research group leadership offers researchers the opportunity to independently lead a research group and to deepen their own research focus. In this context, the scientific independence that the position entails are particularly noteworthy. Institute director Prof. Dr. Christian Scheideler emphasizes the importance and influence on the research profile of Paderborn University: "The newly created junior research group leaders create the opportunity to strengthen and expand the existing research focus and to test new research fields. This also leads to an enrichment of our teaching, which makes studying computer science with us more attractive.".
We wish Dr.-Ing. Philipp Terhörst a good start as a junior research group leader and are very pleased to welcome him to the Institute of Computer Science at Paderborn University.