A0586
Title: Neural statistical modelling of cascading extremes
Authors: Miguel de Carvalho - University of Edinburgh and Universidade de Aveiro (Portugal) [presenting]
Clemente Ferrer - USM (Chile)
Ronny Vallejos - Universidad Tecnica Federico Santa Maria (Chile)
Abstract: The purpose is to address the growing concern surrounding cascading extreme events, such as a major undersea earthquake triggering a tsunami, by presenting a novel method for risk assessment focused on these domino effects. The proposed approach develops an extreme value theory framework within a Kolmogorov-Arnold network (KAN) to estimate the probability of one extreme event triggering another, conditionally on a feature vector. An extra layer is added to the KANs architecture to enforce the definition of the parameter of interest within the unit interval, and the resulting neural model is referred to as KANE (KAN with natural enforcement). The method is supported by extensive numerical studies and further demonstrated through real-world applications. Finally, connections between the proposed methods and ongoing work on generative modeling of extreme events are mentioned.
