Topics will vary from semester to semester. See Computer Science Division announcements.

We will read a modest number of papers on neural networks, with emphasis on their objective functions, robustness, and generalization. This is a research-oriented class (albeit light, hence only 2 units) with a goal of finding some extensions and generalizations of the papers we read. Grades will depend on two things: each student will present one paper (or part of one), and each student should identify several open problems and find at least one extension of the results discussed in class. (These extensions may be very modest for the purposes of grading, but I hope that a few will eventually lead to publications.) For this new, trial course, I will admit Ph.D students only, and only about 15 of them.

* Will NOT allow time conflicts.
* Will only accept PhD students into the class
* All students must join waitlist and professor will select from that list.

Requisites

• Students not in the Master of Engineering Program

Repeat Rules