Theme of this iteration

theoretical foundations of deep learning… puts classic learning theory in the context of modern deep learning, and examines deep learning through the lens of classic and modern approximation, optimization, and learning theory


What this course is NOT about: Introduction to deep learning, or classic machine learning topics covered in CSCI5525. These courses are part of the prerequisites (see below).

Instructor consent only. This is PhD-level special topics course in theoretical/mathematical understanding of deep learning, and targets PhD students with research focus/interest in the foundations of modern machine learning and data science. This course is math-heavy and research-oriented: both course and research experience in advanced machine learning (CSCI5525 or equivalent) and deep learning (CSCI5980: Think Deep Learning or equivalent) are expected, and maturity in linear algebra, multivariate calculus, probability, and numerical optimization is assumed. To get a permission number, please email the instructor and describe how your course and research experience has prepared you for this course.

Course info

Syllabus: TBD
A superset of topics we’d like to cover: Live Google doc (UMN log-in required)

Instructor: Professor Ju Sun Email: jusun AT (Office Hours: TBD)

When/Where: Mon/Wed 2:30 – 3:45pm/TBD