By David Murphy
The KAUST Extreme Statistics (extSTAT) Research Group received two prestigious awards at this year’s Extreme Value Analysis (EVA 2023) Conference in Milan, Italy.
The 13th International Conference on Extreme Value Analysis was held at Bocconi University from June 26 to June 30, 2023. The event brought together leading researchers in extreme value theory (EVT), methods and applications.
KAUST Ph.D. statistics student Paolo Victor Redondo won a Best Poster Award at EVA 2023 for his poster titled "Spectral Transfer Entropy and its Application to Extremal Brain Causal Connectivity Networks." The poster detailed a joint research work with his KAUST co-supervisors, Professor Raphaël Huser (PI of the extSTAT Research Group) and Professor Hernando Ombao (PI of the Biostatistics (BIOSTATS) Group).
The researchers have developed a novel approach to modeling causal relationships between brain regions using EVT. Redondo and his colleagues are using transfer entropy and extreme value statistics to examine brain signals such as electroencephalograms (EEGs).
A key objective of the study is to gain insight into how information flows between brain regions during mental tasks. Their pioneering approach has the advantage of capturing connections linked to brain oscillations associated with cognitive functions.
“For example, connections among low-frequency waves are related to attention and memory, while the execution of voluntary movements is associated with connections among high-frequency waves,” Redondo explained. “By taking the perspective of EVT, our method offers a unique and straightforward solution to address inherent issues in brain signal filtering. And thus, provides an easy-to-interpret tool to visualize dependence in a brain network.”
“Receiving an award of this kind demonstrates that merging various disciplines, such as EVT and modeling cross-dependence in a brain network, paves the way for relevant research opportunities,” he said of his EVA 2023 award win. “Additionally, it serves as a generous reminder that the hard work carried out at KAUST impacts the global stage.”
Team honors for KAUST
The "Yalla Team," composed of Redondo and his extSTAT colleagues Noura Alotaibi, Daniela Cisneros, Yan Gong, Jordan Richards, Xuanjie Shao, and Matheus Bartolo Guerrero, a former KAUST Ph.D. statistics student, now an Assistant Professor of statistics at California State University, Fullerton, U.S., won first place in the EVA 2023 Data Challenge.
The data challenge competition—organized by Dr. Christian Rohrbeck, University of Bath, U.K.; Dr. Emma Simpson, University College London, U.K.; and Professor Jonathan Tawn, Lancaster University, U.K.—consisted of four sub-challenges based on a variety of problems experienced in analyzing environmental extreme data. Given simulated data from unknown complex extreme value models, the tasks included modeling extreme quantiles and estimating tail probabilities in low- and high-dimensional cases.
The KAUST team developed new methodologies to answer each sub-challenge. As a result, they achieved a second-place finish in three out of four sub-challenges and the best performance in the last sub-challenge, which resulted in their overall win in the Data Challenge competition.