Following a brief welcome by Prof. Dr. Jörn Müller-Quade, spokesperson for KASTEL Security Research Labs, two impressive demonstrations on cybersecurity research took place.
First, Alessandro Erba simulated real-world threat scenarios for traffic light recognition in autonomous vehicles. To mitigate these threats, he presented XAI-based defense strategies and anti-backdoor learning techniques as potential solutions.
Jeremias Mechler then demonstrated in a live demo how an AI computation can be executed in a cryptographic enclave on the DKFZ’s Kaapana platform without giving insight into sensitive medical data.
Visitors asked numerous questions, underscoring the strong interest in the presented research results.