Title of presentation
Explaining ZIP Using Information Theory
Biography
Frans M.J. Willems was born in Stein, The Netherlands, in 1954. He received the M.Sc. degree in Electrical Engineering from the Technische Universiteit Eindhoven, Eindhoven, The Netherlands, and the Ph.D. degree from the Katholieke Universiteit Leuven, Leuven, Belgium, in 1979 and 1982 respectively. In 1982 he joined the Electrical Engineering Department of the Technische Universiteit Eindhoven. He is full professor (emeritus) there. His research contributions are in the areas of multi-user information theory, noiseless source coding, data-embedding, and biometrics. From 1999 - 2016 he was an advisor for Philips Research Laboratories, Eindhoven, for topics related to information theory.
Frans Willems received the Marconi Young Scientist Award in 1982. He is co-recipient of the 1996 IEEE Information Theory Society Paper Award for a paper in which the Context-Tree Weighting Algorithm was proposed. From 1998 to 2000 he was a member of the Board of Governors of the IEEE Information Theory Society. Since 2005 Frans Willems is an IEEE Fellow. He was was Chairman of the IEEE Benelux Chapter on Information Theory from 2007 to 2017. Dr. Willems received a 2011 Best Paper Award from the IEEE Signal Processing Society for the paper “Biometric Systems: Privacy and Security Aspects.” In 2014 - 2015 Frans Willems was a Distinguished Lecturer of the IEEE Information Theory Society. He has contributed more than two hundred journal and conference papers. He holds several patents. Recently Frans Willems received the 2024 IEEE Richard W. Hamming Medal.
Title and abstract of presentation
Attacks on Face Biometrics Systems
Face Recognition Systems (FRS) have been widely adopted across various applications, with deep learning advancements significantly improving their accuracy. However, these same advancements—along with the rise of generative AI—have introduced a new wave of sophisticated attacks, making FRS increasingly vulnerable. From Deepfakes and morphing attacks to the more recent adversarial attacks, FRS have repeatedly demonstrated a lack of resilience against evolving threats.
This talk will delve into the attack potential of these techniques, the challenges in detecting such threats, and strategies for enhancing the robustness of FRS. The second half of the discussion will focus on the pressing issues of generalizability and explainability, highlighting their critical role in helping human observers make informed decisions—especially in safety-critical applications.
Biography
Kiran Raja is a faculty member at the Norwegian University of Science and Technology (NTNU), Norway. He has participated in EU-funded projects such as SOTAMD, iMARS, INGRESS and other nationally funded projects on biometrics, security and machine learning. With his participation in SOTAMD and iMARS projects at the Norwegian Biometrics Laboratory (NBL), he works on different problems in morphing attacks from both generation and detection perspectives. In addition, his research interest also lies in detecting manipulations like DeepFakes and adversarial attacks. He is a member of the European Association of Biometrics (EAB) and chairs the Academic Special Interest Group at EAB. He also serves as program co-chair for the BIOSIG conference. He also advises various national agencies in Norway on making biometric systems secure. His recent research focuses on attacks and defences on biometric systems using pattern recognition and machine learning approaches.