Martin Clyde, PhD Professor Department of Mathematics and Statistics Texas Tech University
Abstract: Smoothing splines are a powerful tool for representing data as long as the data has enough nice properties. In this talk I will discuss three problems and what we are doing to solve them.
1)1) L1 splines rather than L2 splines useful when the data is not normal. Work with Masaki Nagahara.
2)2) The underlying state space is not Euclidian. Work with Wenzhen Fan and Jingyong Su.
3)3) Very large data sets such as those encountered in tracking eye and head movement. Work with Bijoy Ghosh and Jennifer Emerson.
Problem 1 arises when the data has many outliers such as climate data. L1 splines were much more efficient than L2 splines. Problem 2 arose in a certain tracking problem that was best stated as a problem on the manifold of lines in R3. It also arose in considering a certain climate data and was reduced to the manifold of intervals. Problem 3 arose in the study of head and eye movement. Data sets of 300000 to 600000 points were used.
Title and abstract: TBD
Ming Yang, Ph.D. Assistant Professor Division of Biostatistics and Bioinformatics Penn State University