Today, the Department of Statistics at Iowa State University hosted Christopher K. Wikle, Curators’ Distinguished Professor and Chair from the University of Missouri, for a seminar on "Flexible and Efficient Spatial Extremes Estimation and Emulation via Variational Autoencoders." Wikle discussed the challenge of assessing extreme exposures, emphasizing the limitations of current methodologies in capturing complex tail dependence structures in spatially dependent processes.
Wikle introduced a novel spatial extremes model integrated into the encoding-decoding structure of a variational autoencoder (XVAE). This approach, utilizing variational Bayes combined with deep learning techniques, offers flexibility and efficiency in analyzing high-dimensional data and emulating spatio-temporal distributions. Through simulation studies, Wikle demonstrated that the XVAE is more time-efficient than traditional Bayesian inference methods, particularly in capturing extreme events.
Attendees gained insights into the application of these techniques to real-world datasets, such as high-resolution satellite-derived sea surface temperature data and extreme wildfire simulations. Wikle highlighted the versatility of the proposed methods, suggesting their potential application to various datasets exhibiting extremes, including those in public health contexts.
Overall, the seminar provided a glimpse into cutting-edge methodologies for spatial extremes estimation, showcasing the potential of variational autoencoders to enhance efficiency and accuracy in analyzing extreme events. Wikle's presentation opened avenues for further exploration and application of these techniques across diverse domains within statistics and beyond.
