Graph-based Learning and Analysis for intrusion Detection in Information Systems – GLADIS

GLADIS project aims to build an efficient real-time cyber-attack tracking system based on a graph representation of the system activities. The main scientific novelty consists in modelling heterogeneous logs incoming from different devices into a set of dynamic graph structures in order to track the different activities and behaviors, pinpoint the abnormal ones and trace the sources of the attacks using graph analytics techniques (graph learning and anomaly detection). To meet the objectives of the project, two teams with complementary skills are involved: SnT provides its knowledge and experience of cybersecurity attacks and LIRIS provides, through the Graph, algorithms and applications team, its competencies on graph theory and applications.