Activities

Reading Group

The reading group consists of monthly two-hour online meetings. Each meeting will consist of two halves: a discussion of a recent piece of contributing literature followed by the opportunity for one of our members to present ongoing research or brainstorm new ideas. There will be 10-minute breaks between each part.

Click here to see the Zoom link to our next meeting.

Meeting structure

  • First half: during the first 55 minutes of each session, we cover a paper. There will be a lead discussant who sets the discussion in motion, and then the group jumps in.
  • In the last final minutes, we wrap up and look for a volunteer to lead the next session (a new discussant may be recommended by PhD supervisors).

  • A 10-minute break.

  • Second half: the last 55 minutes are for a research presentation or brainstorming new ideas.
  • In the final minutes, we look for a volunteer to present their research/discuss ideas in the next session (again, may be recommended by PhD supervisors).

 

Reading list

Click here to download the articles.

Date and timeArticleDiscussant
18/10/24 13:00-14:00
Olafsdottir, H. K., Rootzén, H., & Bolin, D. (2021). Extreme rainfall events in the Northeastern United States become more frequent with rising temperatures, but their intensity distribution remains stable. Journal of Climate, 34(22), 8863-8877.
 Daniela Castro-Camilo
15/11/24 13:00-14:00
Li, R., Leng, C., & You, J. (2020). Semiparametric Tail Index Regression. Journal of Business & Economic Statistics, 40(1), 82–95.
 Johnny Lee

 

Research presentation/brainstorming

Friday 15th November 14:00-15:00

Title: A Kolmogorov–Arnold Neural Model for Cascading Extremes

By: Miguel de Carvalho

Summary

TIn this talk I will address the growing concern of cascading extreme events, such as a tsunami followed by an extreme earthquake, by presenting a novel method for risk assessment focused on these domino effects. The proposed method develops an extreme value theory framework within a Kolmogorov–Arnold Neural Network (KAN) to estimate the probability of one extreme event triggering another, as a function of a covariate or feature vector. Our approach is backed by exhaustive numerical studies and illustrated on a real-life application to seismology.
 
Joint work with C. Ferrer and R. Vallejos.
 

Friday 18th October 14:00-15:00

Title: A deep learning approach to modelling joint environmental extremes

By: Jordan Richards

Summary

The geometric representation for multivariate extremes, where data is split into radial and angular components and the radial component is modelled conditionally on the angle, provides an exciting new approach to modelling environmental data. Through a consideration of scaled sample clouds and limit sets, it provides a flexible, semi-parametric model for extremes that connects multiple classical extremal dependence measures; these include the coefficients of tail dependence and asymptotic independence, and parameters of the conditional extremes framework. Although the geometric approach is becoming an increasingly popular modelling tool for environmental data, its inference is limited to a low dimensional setting (d ≤ 3). 
 
We propose here the first deep representation for geometric extremes. By leveraging the predictive power and computational scalability of neural networks, we construct asymptotically-justified yet flexible semi-parametric models for extremal dependence of high-dimensional data. We showcase the efficacy of our deep approach by modelling the complex extremal dependence between metocean variables sampled from the North Sea.
 
Joint work with Callum JR Murphy-Barltrop and Reetam Majumder
 

Workshops

Two half-day hands-on workshops will be held each year, shaped by the reading group discussions. More detailed information will appear here in due course.