New Publication: Towards Formal Security Analysis of Deep Neural Network Architectures
Our paper,”Towards Formal Security Analysis of Deep Neural Network Architectures,” has been published in Innovations in Systems and Software Engineering.
Neural networks are increasingly used in security-critical domains, yet their susceptibility to attacks and manipulation remains a key challenge. In this work, we investigate the use of formal techniques to verify security-related properties of deep neural network architectures, contributing toward more rigorous assurance frameworks for AI-enabled systems.
This is joint work with researchers from IRIT and CEA List.
See Publications for more details!