Plenary Talk

Title: High-dimensional Data Analysis
Presenter: Professor Zhidong Bai, National University of Singapore, Singapore
Abstract:

With the rapid development of computer science and its wide application, people are able to collect, store and deal with huge amount of computation with data sets of large sizes and large dimensions. Besides the advantages the new development brought to us, a lot of challenges have brought to us also. The first question is whether the classical limit theorems (i.e., those developed under the assumption that the dimensional is fixed) and statistical inference methods developed based on them are still valid for large dimensional data analysis? This has been confirmed in the literature. And thus, many have proposed the Large dimensional data analysis. But few know what it is. I think, the most promising way is to borrow the spectral theory of large dimensional random matrix to the statistical analysis of large dimensional data.

In the first talk, I would introduce some examples to show the difference between large and small dimensional data analysis. How serious classical limiting theorems made errors in statistical inferences on large dimensional data.