Course Description

This is a seminar course in statistical learning theory. Statistical learning theory is a widely popular framework for studying the problem of inference--making predicitons, making decisions, or constructing models from a set of data. This course will cover a small subset of seminal papers in statistical learning theory, which will be discussed by members of the class.

"Nothing is more practical than a good theory." - V. Vapnik

Course Organizers

Daniel Spokoyny
Kevin Malta

Faculty Sponsor

Ömer Egecioglu

Class Time and Location

Spring quarter (March - June, 2016).
Lecture: Tuesday 4:00-5:15
Phelps, 3526

Office Hours

TBD!

Contact Info

Daniel: daniel@spokoyny.me
Kevin: kmalta@umail.ucsb.edu
Omer: omer@cs.ucsb.edu

Schedule

*Subject to Change

Event TypeDateDescriptionCourse Materials
Seminar Mar 29 Logistics
Read the paper in the link on the right, the video and Week 8's video lectures you can find by following the remaining links on the right. The videos are optional but may be helpful for those new to the subject.
Seminar Apr 5 Statistical Learning Theory Overview [Paper]
[Video]
[Intro Lectures]
Seminar Apr 12 A Theory of the Learnable [Paper]
Seminar Apr 19 SVMs
The second link provides a helpful intuitive understanding of SVMs, highly recommended.		 [Paper]
[SVM Tutorial]
Seminar Apr 26 Bagging Predictors [Paper]
Seminar May 3 AdaBoost/Boosting [Paper]
Seminar May 10 MCMC/Metropolis-Hastings [Paper]
Seminar May 17 Maximum Likelihood/EM Algorithm [Paper]