ECE 543: Statistical Learning Theory (Spring 2023)

ECE 534 (Random Processes)

The course will be based on the following sets of notes:

Additional References

S. Shalev-Shwartz and S. Ben David, Understanding Machine Learning: From Theory to Algorithms
M. Mohri, A. Rostamizadeh, and A. Talwalkar, Foundations of Machine Learning
Vladimir N. Vapnik, Statistical Learning Theory
Trevor Hastie, Robert Tibshirani, Jerome Friedman, The Elements of Statistical Learning: Data Mining,
Inference, and Prediction

Homework (60%), Scribing (10%), Final Project (30%)

Homework: You can discuss the homework problems in general terms. Nevertheless, identical solutions (copying)
will not be allowed. In other words, you are expected to work on your own solutions. The solutions will be submitted
on gradescope. Late homework will not be accepted.
Final Project: Similarly to Spring 2021 (see the course webpage by Prof. Raginsky below), you may select and evaluate
1 paper relevant to the course material from conferences such as COLT, ICML, ALT, AISTATS and NeurIPS. Moreover,
you should work in groups of preferably 4 students. A brief project proposal from each group is due by Mon, April 10.

Project Presentations: Each group will give a presentation based on the selected paper in the end
of the semester. Students in each group will assume one of the following different “roles”:
-Presenter: Create the main presentation, describing the motivation, problem definition and results in the selected
paper.
-Reviewer: Complete a full critical review of the selected paper by following the guidelines for NeurIPS reviewers
(under “Review Content”) and assign an Overall score and a Confidence score. You are not required to present the
related work.
-Archaeologist: Determine the position of the selected paper in the context of previous and subsequent work.
Find and report on one older paper that has substantially influenced the selected paper and one newer paper citing
this paper.
-Industry Expert: Propose a new application or a product based on the results of the presented paper and discuss
positive and negative impacts of this application. Convince your industry collaborators that it is worth investing time
and money to implement this product by relying on arguments applicable to the particular industry market.
NOTE: You are not expected to put effort on deriving new results or extensions based on the selected
paper. To control the course workload and to also ensure a concrete and streamlined framework for
the grading of your final projects, you should restrict your presentations only to the results appearing
in the selected paper by your groups.

Presentation Roles: The presentation roles are based on Colin Raffel’s COMP790 course at the University of North
Carolina, Chapel Hill and Alec Jacobson’s CSC2521 course at the University of Toronto.

Homework submissions via gradescope

Lecture Schedule and Homework Problems

Please check past offerings of the course by Profs. Hajek, Srikant and Raginsky: