Joint ECE and DSC Seminar: Scalable Spatio-Temporal Regression and Deep Neural Network
Modern scientific studies often generate massive spatial and spatio-temporal datasets, creating new challenges for statistical modeling and computation. In this talk, I will discuss recent work on scalable methods for analyzing such data, with applications motivated by environmental and geospatial studies. The first part of the talk introduces a regression framework for large spatio-temporal datasets that accounts for spatial and temporal dependence while remaining computationally efficient for large-scale inference. The second part presents a deep learning approach for spatial regression that can capture complex nonlinear relationships between outcomes and covariates. Together, these methods illustrate how modern statistics and machine learning can be combined to address important challenges in large-scale spatial data analysis.
SENG 213A