STAT 593C:
Model Selection and Regularization
Spring Quarter 2007
Syllabus (last updated: 2/26/07)
Course personnel:
- Professor: Jon A. Wellner
- B320 Padelford Hall
- Phone: 543-6207
- Office hours: 1:30 - 3:30 MWF; or by appointment
Time and Place:
- Time(s): T Th 1:30 - 3:00
- Place: Miller 302B
Prerequisites:
- Successful completion of STAT 582
- Probability theory, ideally at the level of STAT 521.
Textbooks (not required):
- Trevor Hastie, Robert Tibshirani, and Jerome Friedman.
The Elements of Statistical Learning.
Springer-Verlag, New York.
- Pascal Massart.
Concentration inequalities and model selection. (2006).
Ecole d'Ete de Probabilites de Saint-Flour XXXIII-2003.
Springer, Berlin.
Course Description:
This special topics course will treat methods for model selection
and regularization, including Akaike's AIC, Mallows' C_p, Schwarz's BIC,
generalized AIC, cross-validation and generalized cross-validation.
The emphasis will be on classification and regression problems, but
other statistical settings will also appear.
We will relate these to other alternative methods of regularization
including ridge regression, lasso methods, and various randomized
penalty terms. We will describe asymptotic
approaches to theoretical understanding of the methods as well as
finite-sample oracle inequalities making use of empirical process techniques
due to Talagrand and developed further by Birge, Massart, and Koltchinskii.
Click here to return to Jon Wellner's home page.
Click here
to return to the Statistics Dept. home page.
Click here
to return to the University of Washington home page.