A Survey on Robust Low-Rank Matrix Estimation

PhD Qualifying Examination


Title: "A Survey on Robust Low-Rank Matrix Estimation"

by

Mr. Naiyan WANG


Abstract:

Robust low-rank matrix estimation has emerged as a hot research topic in 
recent years.  This important problem underlies many methods that have 
been found highly effective in revealing the low-dimensional subspace 
structures of complicated high-dimensional data, especially when the data 
matrix is corrupted by outliers or missing entries.  With advances in both 
the theory and specific optimization techniques, a wide range of novel 
applications have been proposed by the computer vision and machine 
learning communities.

In this paper, we survey the major approaches to robust low-rank matrix 
estimation.  We first start with different formulations of the problem. 
Then we categorize them into optimization-based and Bayesian methods and 
discuss their relationships and differences.  After that, we briefly 
discuss some representative applications which take effective use of 
robust low-rank matrix estimation.  Finally, we conclude with recent 
development of this topic and discuss several possible future research 
directions.


Date:			Friday, 19 April 2013

Time:	 		2:00pm - 4:00pm

Venue:   		Room 3401
 		        Lifts 17/18

Committee Members:	Prof. Dit-Yan Yeung (Supervisor)
  			Prof. James Kwok (Chairperson)
  			Dr. Albert Chung
  			Prof. Nevin Zhang


**** ALL are Welcome ****