The Foundations, Algorithms, and Applications of Non-Negative Matrix Factorization

PhD Thesis Proposal Defence


Title: "The Foundations, Algorithms, and Applications of Non-Negative Matrix 
Factorization"

by

Miss Qing LIAO


Abstract:

In recent years, the parts-based representation has been shown a powerful 
representation tool for various practical applications in machine learning and 
data mining because it is consistent with the psychological and physical 
evidence in human brain. Non-negative matrix factorization (NMF) is a dimension 
reduction method which decomposes a given non-negative data matrix into the 
product of two lower-rank non-negative factor matrices, i.e., the bases and the 
coefficients. Due to its simplicity and effectiveness, NMF has been extended to 
meet the requirements of various applications, e.g, clustering and 
classification.

In this proposal, we first introduce the background knowledge and properties 
about NMF in the following aspects, i.e., models, algorithms, and applications. 
To overcome the deficiencies of NMF or to meet the requirements of 
applications, we design several NMF extensions including Logdet divergence 
based sparse NMF (LDS-NMF), robust local coordinate NMF (RLC-NMF), and local 
coordinate graph regularized NMF (LCG-NMF). To accelerate the optimization of 
NMF, we develop a novel algorithm (RRA) for optimizing NMF. Finally, we apply 
NMF models to solve practical problems in the application part.


Date:			Friday, 8 April 2016

Time:                  	1:15pm - 3:15pm

Venue:                  Room 3588
                         (lifts 27/28)

Committee Members:	Prof. Qian Zhang (Supervisor)
  			Dr. Wei Wang (Chairperson)
 			Dr. Lei Chen
  			Dr. Qiong Luo


**** ALL are Welcome ****