550:640 Machine Learning

Spring 2009

Whitehead 304, 3:00-4:15pm

 

 

 

Laurent Younes

Department of Applied Mathematics and Statistics

Center for Imaging Science

The Johns Hopkins University

3400 N Charles Street

21218 Baltimore MD

 

Clark 324C

 

Email:   (replace * with dots):  laurent*younes at jhu*edu

 

 

 

Office hours: By appointment.

 

Textbook: The class follows the handouts available below.

 

Content of the course (probably a superset):

bullet Introduction to density estimation.
bullet Linear methods for regression and classification (regression, ridge regression, lasso, linear support vector machines; logistic regression, linear discriminant analysis)
bullet The notion of feature space; an introduction to kernels; kernel SVM
bullet Nearest neighbor methods
bullet Introduction to model complexity
bullet Boosting
bullet Trees and random forests
bullet Unsupervised methods: (kernel) PCA and variants, multidimensional scaling, local linear embedding, isomap, independent component analysis
bullet Clustering

Handouts (password protected):  Introduction to Machine Learning.

Projects (password protected):

bullet

project 1 (additional file project1.dat)

bullet

project 2 (additional file project2.zip)

bullet

project 3 (additional file project3.zip)

bullet

project 4 (additional file USPS.zip)

bullet

project 5 (additional file project5.zip)

 

   

Links to a few packages:

bullet A matlab SVM (A. Schwaighofer)
bullet SVM-light
bullet Tom Briggs's wrapper to SVM-light, with precompiled libraries here
bullet An R package containing SVM's