Séminaire de recherche Opérations - "FAnalyzing Contagious Cyber Incidents and Extreme Climate Events with Point Process Models" - Antoine Heranval, INRAE Avignon

In the first part of this presentation, we introduce a method to classify sectors based on their vulnerability to contagious cyber incidents. By integrating Multivariate Hawkes processes with the Classification and Regression Trees (CART) algorithm, we capture the temporal dynamics and interconnections of these events. Using the likelihood of the Hawkes process as a splitting criterion, the approach emphasizes incident frequency rather than severity. Results are validated through simulations and real-world cyber incident data, offering practical insights for risk management. The second part examines dependencies between different types of extreme climate events and the occurrence of compound events. Each event (e.g., temperature, precipitation, wind speed) is characterized by its time of occurrence and attributes such as magnitude, duration, or spatial extent. Using real-world climate datasets, we analyze statistical features like intensity functions and correlation measures to study how events interact across scales. We also propose models on spatial graphs to represent these dependencies, enabling the simulation and analysis of compound event across regions.