Welcome at the Chair of Privacy and Security
We work on all aspects of technical data protection and network and IT security. We are primarily interested in privacy, i.e. the protection of individuals against the misuse of their data.
Over the past decades, digital technologies have developed rapidly. The advent of digital transformation and the interconnectivity of all areas of life opens up a wealth of new possibilities. Autonomous networked vehicles, cloud computing, industry 4.0, virtual reality with haptic feedback, online banking or social networks are just a few keywords that are changing the way and quality of life. However, this development also brings with it a number of challenges that people often do not immediately grasp, but which may well conflict with their interests. Our research group is engaged in the development and analysis of security concepts that protect from all types of potential attackers on such systems. We are also interested in the development of technologies that promote data protection in order to protect privacy in the digital world. Finally, we develop protocols and algorithms to secure the underlying infrastructures for communication and computation.
We are part of the KASTEL Security Research Labs, as well as the excellence cluster CeTI, the Centre for Tactile Internet with Human-in-the-Loop.
The paper A kinematic dataset of locomotion with gait and sit-to-stand movements of young adults
(Simon Hanisch, Loreen Pogrzeba, Evelyn Muschter, Shu-Chen Li, Thorsten Strufe) has been published in Nature Scientific Data. In this paper we present a new motion dataset in which we recorded 50 young adults performing different gait exercises and a sit-to-stand exercise. The dataset was recorded using IMU-suits and will allow new research into anonymizing motion data.
Congratulations!
Link to the paperThe paper R+R: Understanding Hyperparameter Effects in DP-SGD
(Felix Morsbach, Jan Reubold, Thorsten Strufe) has been accepted at ACSAC 2024. In this work, we examine the influence that the hyperparameters have on the learning process in privacy-preserving machine learning.
Congratulations!
Link to the paperOn the 25th of October, Prof. Joel Reardon from the University of Calgary will give a talk titled “Anonymity, Consent and Other Noble Lies: An Empirical Study of the Data Economy”. In this talk, he will tell us about empirical studies with data acquired from data brokers, and how the anonymization methods they employ are not enough. In their result, they show that people can be re-identified, and that data has been collected without proper consent from the individuals.
We are looking forward to an exciting talk!
Our Patricia Guerra Balboa was at the Workshop on Privacy in the Electronic Society (WPES) in Salt Lake City. There she has presented her work on resilience bounds of differential privacy in the context of membership inference, attribute inference and data reconstruction attacks, work which has been done together with Annika Sauer. The paper also introduces a general, unifying model with newer, much tighter bounds.
