Instructor: Bo Waggoner

Course webpage: https://www.bowaggoner.com/courses/2021/learning-theory/

Presents the underlying theory behind machine learning in proofs-based format. Answers fundamental questions about what learning means and what can be learned via formal models of statistical learning theory. Analyzes some important classes of machine learning methods. Specific topics may include the PAC framework, VC-dimension and Rademacher complexity.

Undergraduate analysis (APPM 4440 “Real Analysis” or equivalent, or even better, graduate analysis like APPM 5440) and mathematical maturity (and we assume the usual prereqs for analysis, such as linear algebra and probability), as this is a proofs-based math class. Familiarity with machine learning algorithms is very helpful (e.g., CSCI 5622).

The course will be primarily lecture-based, with regular homework and a course project. It will meet virtually, but synchronously ("remote" modality).

This is a lecture-based course, and the instructor will present proofs. The textbooks provide supplementary material and homework problems. Students are expected to read the book as necessary to fill in gaps not covered in lecture. Reading the book before lecture is useful but not required. Homeworks will require students to write proofs, synthesizing concepts learned from lecture and the book.

- Homework (70%): Assignments due about once every two weeks.
- Final project (30%): In groups of one or two, students will investigate a topic related to learning theory. They will turn in a project proposal of about two pages, biweekly progress reports of a couple paragraphs, and a final writeup of 8-15 pages; and give a 10-30min in-class presentation on the project.

The final score will be calculated by a weighted average of the grades in each component. Course letter grades will be assigned based on the final score.

Each student's final homework score will be the average of their percentage scores on all but their worst two assignments.

A student experiencing unusual circumstances preventing them from turning in a homework may simply skip that assignment and take a zero; this will be one of the dropped grades.

Therefore, there will be **no late or make-up assignments**. A student experiencing circumstances likely to cause three or more missed homework should reach out to the instructor to discuss whether it is possible to complete the course given this disruption.

Standard CU course policies can be found here: https://www.colorado.edu/academicaffairs/policies-customs-guidelines/required-syllabus-statements.