Haijun Li's Analysis Seminar Talks for March 2 and 9, 2011
Title: Regularly Varying Random Sets
Abstract: We will discuss the ideas to develop a geometric extreme value theory for random sets with regularly varying boundaries. In the first part of the talk, we will discuss the motivation and need for such a geometric theory, and show why the traditional, coordinate-wise multivariate extreme value theory becomes cumbersome in analyzing extremes in high-dimensional data analysis. In the second part of the talk, we will discuss weak convergence of properly scaled unions of random sets and highlight open questions. We will introduce all the basic ingredients in understanding the essential ideas, and in particular, no prior knowledge on random sets is required for this exploratory talk.
Background:
