High-dimensional canonical correlation analysis

Time：2023-10-25, Wednesday, 16:00-17:00

Venue：C654, Shuangqing Complex Building

Organizer：吴昊，杨帆，姜建平，顾陈琳

Speaker：Vadim Gorin

Speaker：

Vadim Gorin

Associate Professor of Statistics and Mathematics at University of California, Berkeley

Abstract：

Canonical correlations have two faces: from one side, in statistics they give a way to measure dependence between two datasets. From the other side, in linear algebra they represent a canonical form to which a pair of two linear subspaces can be transformed. In the talk we will discuss the theory of canonical correlations when the dimensions of the subspaces are large, in the setting related to both random matrix theory and high-dimensional statistics.

DATEOctober 25, 2023

Related News

0
Canonical colourings in random graphs | Research seminar in Discrete Mathematics
AbstractRdl and Rucinski have extended Ramseys Theorem to random graphs, showing that there is aconstant C' such that with high probability, any two-colouring of the edges of G(n, p) with edgeprobability p= C'n?/(t+1) contains a monochromatic copy of Kt (the complete t-vertex graph). Weinvestigate how this statement extends to arbitrary colourings of G(n,p). Namely, when noassumptions are made ...
1
Manifold learning for noisy and high-dimensional datasets: challenges and some solutions
Abstract：Manifold learning theory has garnered considerable attention in the modeling of expansive biomedical datasets, showcasing its ability to capture data essence more effectively than traditional linear methodologies. Nevertheless, prevalent algorithms are primarily designed for low-dimensional and clean datasets, whereas contemporary biomedical datasets tend to be high-dimensional and no...