Online Seminar Series
GAMEX hosts an online seminar series throughout 2026 to stimulate scientific exchange within the network and beyond, with the support of the Glasgow–Edinburgh Extremes Network (GLE²N) and the Community of Italian Researchers Connecting on Extreme Value Statistics (CIRCE).
Program
April 10, 2026
Time: 9:30 ET (13:30 UTC / 14:30 BST / 15:30 CEST)
Speaker: Frédéric Godin (Concordia University)
Title:
Catastrophic-risk-aware reinforcement learning with extreme-value-theory-based policy gradients
Abstract:
We tackle the problem of mitigating catastrophic risk (which is risk with very low frequency but very high severity) in the context of a sequential decision making process. This problem is particularly challenging due to the scarcity of observations in the far tail of the distribution of cumulative costs (negative rewards). A policy gradient algorithm is developed, that we call POTPG. It is based on approximations of the tail risk derived from extreme value theory. Numerical experiments highlight the out-performance of our method over common benchmarks relying on the empirical distribution. An application to financial risk management, more precisely to the dynamic hedging of a financial option, is presented.
Access link:
Join the seminar on Zoom
Passcode: yRkNi2kr
--
May 15, 2026
Time: 9:00 CT (14:00 UTC / 15:00 BST / 16:00 CEST)
Speaker: Likun Zhang (University of Missouri)
Title:
Modeling Spatio-temporal Extremes via Conditional Variational Autoencoders
Abstract:
Extreme weather events are widely studied in fields such as agriculture, ecology, and meteorology. The spatio-temporal co-occurrence of extreme events can strengthen or weaken under changing climate conditions. In this paper, we propose a novel approach to model spatio-temporal extremes by integrating climate indices via a conditional variational autoencoder (cXVAE). A convolutional neural network (CNN) is embedded in the decoder to convolve climatological indices with the spatial dependence within the latent space, thereby allowing the decoder to depend on the climate variables.
There are three main contributions. First, through extensive simulations, we show that the proposed conditional XVAE accurately emulates spatial fields and recovers spatially and temporally varying extremal dependence with very low computational cost after training. Second, we provide a simple and scalable approach to detecting condition-driven shifts and assessing whether the dependence structure is invariant to the conditioning variable. Third, when dependence is condition-sensitive, the conditional XVAE supports counterfactual experiments by intervening on the climate covariate and propagating the change through the learned decoder to quantify differences in joint tail risk, co-occurrence ranges, and return metrics.
We illustrate the methodology by analyzing the monthly maximum Fire Weather Index (FWI) over eastern Australia (2014–2024), conditioned on the El Niño/Southern Oscillation (ENSO) index.
Access link: Join the seminar on Zoom
--
October 2, 2026
Time: 15:30 CEST (13:30 UTC / 14:30 BST)
Speaker: Antonio Ocello (ENSAE Paris)
Title:
Do Heavy Tails Help Diffusion? On the Subtle Trade-off Between Initialization and Training
Abstract:
Recent works have proposed incorporating heavy-tailed (HT) noise into diffusion- and flow-based generative models, with the goals of better recovering the tails of target distributions and improving generative diversity. This motivation is intuitive: if the data are heavy-tailed, HT noise may appear better matched than light-tailed (LT) Gaussian noise. However, replacing Gaussian noise by HT noise also changes the underlying estimation problem.
In this talk, we revisit this paradigm through a combined theoretical and empirical study, establishing sampling-error bounds for two representative diffusion models driven by HT and LT noise. We show that HT noise makes the statistical estimation problem harder, leading to less favorable sampling-error bounds. We support these findings with experiments on synthetic and real-world datasets, empirically recovering the predicted error trade-off.
Our results call into question a growing design trend in generative modeling and challenge the use of HT noise to improve rare-region exploration.
This talk is based on joint work with Hamza Cherkaoui and Hélène Halconruy:
arXiv:2605.13175
Connection details:
Join the seminar on Microsoft Teams
Meeting ID: 349 080 320 615 578
Passcode: ZL64qq2j
December 4, 2026
Time: 15:30 CET (14:30 UTC / 14:30 GMT)
Speaker: Caroline Cognot (UCLouvain)
Title:
TBC
Abstract:
TBC
Connection details:
Join the seminar on Microsoft Teams
Meeting ID: 349 080 320 615 578
Passcode: ZL64qq2j
Propose a seminar
To propose a seminar, please contact the seminar coordinators with:
- Name and affiliation
- Tentative title and short abstract
- Preferred time window
Contact details are available on the Contact page.