CSE 494, CSE/CBS 598: Numerical Linear Algebra for Data Exploration

Lecture Hours (Fall 2007): 10:40am—11:55am (BYAC 110)


Dr. Jieping Ye



Office hours:

MW 2:30pm—4:00pm






Linear Algebra has contributed many methods for handling very large quantities of numerical data. This course will cover linear algebra techniques useful in data exploration. Topics include least squares, matrix decompositions, tensor decompositions, eigenvalue and generalized eigenvalue decompositions, and their applications to data mining, machine learning, computer vision, information retrieval, bioinformatics, and computer graphics. Both theoretical and algorithmic aspects are considered.


Prerequisite: Basic linear algebra skills.


Textbook: Matrix Methods in Data Mining and Pattern Recognition, by Lars Elden, SIAM, 2007.


Topics to be covered:

·         Linear algebra background

o        Vectors and Matrices

o        Linear Systems and Least Squares

o        Singular Value Decomposition

o        Reduced Rank Least Squares Models

o        Tensor Decomposition

o        Clustering and Non-Negative Matrix Factorization

·         Applications

o        Classification of Handwritten Digits and face images

o        Text Mining

o        Page Ranking for a Web Search Engine

o        Automatic Key Word and Key Sentence Extraction

o        Massive data compression in computer graphics using tensor SVD

o        Clustering and classification of Microarray gene expression data

o        Gene expression pattern image classification and retrieval


Course Materials: