CS724 Topics in Algorithm Theory and Design: Spring 2014

Professor Dan Simovici

 

This incarnation of cs724 is dedicated to a presentation of linear algebra algorithms in machine learning and data mining.  This course requires a background in linear algebra and will discuss such topics as spectral clustering, principal component analysis, latent semantic indexing, shape recognition algorithms, applications in classification, etc. This will be supplemented by a presentation of several special linear algebra topics that you did not see in the typical linear algebra course, such as various types of matrix decompositions, singular values, applications of linear algebra in graph and network theory,  spectral graph theory.  Taking this course may help significantly with your research work; for this reason, the course is intended to attract mainly doctoral students.

Here is a list of topics that I intend to cover:

Linear spaces and matrices

Norms and scalar products

Spectral theory for matrices

Linear algorithms in graphs

Principal component analysis

Latent semantic indexing

Biplots

Linear regression

Support vector machines

 

 

The main reference for this are several handouts that will be posted on this page.  We will use Matlab or R as auxiliary software.

Handouts will be posted here:

Spectral Properties of Matrices (slides)

Introduction to R

 

Homeworks will be posted here:

Homework 4

 

Grades in this course will be based on homework, a take-home examination, class participation.