In many real-world applications such as intelligent transportation, biosurveillance, climate science, and bioinformatics, statistical models and machine learning algorithms are applied to large-scale spatio-temporal data. Such data are typically high-dimensional and demonstrate complex nonlinear spatial and temporal dependencies. In recent years, graph machine learning (GML) has emerged as a powerful machinery to harness the rich information encoded in various spatio-temporal datasets. However, the existing models still tend to be insufficient to handle complex structural phenomena exhibited by spatio-temporal processes and do not explicitly account for time-conditioned properties of the encoded knowledge. In this talk, I will demonstrate how our innovative approaches harnessing the interdisciplinary strengths of topological data analysis, statistics, and machine learning allow us to tackle these limitations across a spectrum of spatio-temporal applications. In particular, I will focus on topology-guided deep learning models for spatio-temporal processes that incorporate time-aware shape descriptors and discuss how pushing forward the performance boundary of GML models can assist in decision-making under uncertainties. I will showcase the applications of our models to challenging problems in spatio-temporal forecasting and dynamic link prediction tasks.
Dr. Yuzhou Chen is a tenure-track Assistant Professor in the Department of Statistics at University of California, Riverside. He is also an adjunct professor in the Department of Computer and Information Sciences at Temple University and a Visiting Research Collaborator in Department of Electrical and Computer Engineering at Princeton University. Before that, Dr. Chen worked as a postdoctoral scholar in the Department of Electrical and Computer Engineering at Princeton University. Dr. Chen received his Ph.D. in Statistics from Southern Methodist University. His research focuses on geometric deep learning, topological data analysis, knowledge discovery in graphs and spatio-temporal data, with applications to power systems, biosurveillance and environmental data analytics. His research has appeared in the top machine learning and data mining top conferences, including ICML, ICLR, NeurIPS, KDD, AAAI, ICDM, ECML-PKDD, etc. He was the recipient of 2024 American Statistical Association on Joint Statistical Computing and Statistical Graphics Section Best Student Paper Award, 2021/2022 American Statistical Association Section on Statistics in Defense and National Security Best Student Paper Award, and 2021 Chateaubriand Fellowship from the Embassy of France in the United States.