The Texas A&M Institute of Data Science (TAMIDS) invites you to attend the 2022 TAMIDS Data Science Webinar series on Apr 19th, 2022. This webinar series will introduce some intermediate and advanced topics of Data Science to Texas A&M students and researchers.
Presenter:
Dr. Jian Cao joined the Department of Statistics, College of Science, in June 2021 as a Postdoctoral Researcher with Dr. Matthias Katzfuss, Associate Professor, supported in part under the TAMIDS Postdoc Project Program. Dr. Cao obtained his Ph.D. degree in Statistics from King Abdullah University of Science and Technology (KAUST) in 2020. In 2018, he won the Student Paper Award from the Statistics Computing Section of the American Statistics Association. Dr. Cao’s main research interests include computational statistics, Gaussian processes, and parallel computing. For his TAMIDS activities, Dr. Cao collaborates with Dr. Katzfuss and Dr. Seth Murray, Professor and Butler Chair for Corn Breeding and Genetics, Department of Soil and Crop Sciences, on statistical modeling and prediction for corn-field growth.
Time:
2:00 PM – 3:30 PM CDT, Apr 19th
Description:
Gaussian process (GP) is a non-parametric model that has good analytical properties, commonly used in Bayesian learning. It is capable of both prediction and uncertainty quantification. Direct training of GP requires order-three complexity but various GP approximations are proposed to reduce the complexity to linear while maintaining a decent accuracy. In this tutorial, we first cover the concept of GP and fit it into a weight-height dataset. Then we introduce a popular linear GP approximation, the Vecchia approximation, and highlight the speed improvement of fitting the same dataset.
Prerequisites:
Basic knowledge of linear algebra, calculus, statistics