Computer Science 692K - S- Algorithms with Predictions

This seminar course will cover the recent topics in the emerging field of Algorithms with predictions, also known as learning-augmented algorithms, or algorithms with ML advice. This is an emerging topic at the intersection of theoretical computer science and machine learning. Generally speaking, a result in this area takes a problem with strong information-theoretic lower bounds (for instance on the competitive ratio), identifies a compact prediction that can be learned from real data, and gives a proof tying the performance of the algorithm to the quality of the underlying prediction. The field has blossomed with applications to a broad range of classic domains in theoretical computer science such as classical streaming algorithms, online scheduling, clustering, filtering data structures, and many others. The focus of this seminar, however, will be on online algorithms with predictions.

Open to Graduate Computer Science students only. SECTION 01=3 CREDITS; SECTION 02=1 CREDIT. A PROJECT IS REQUIRED FOR STUDENTS TAKING FOR 3 CREDITS. THIS IS A THEORETICAL SEMINAR COURSE WITH AN EMPHASIS ON ALGORITHM DESIGN, CORRECTNESS PROOFS, AND ANALYSIS. A GENERAL BACKGROUND IN ALGORITHMS AND A STRONG MATHEMATICAL BACKGROUND ARE REQUIRED. STUDENTS NEEDING SPECIAL PERMISSION MUST REQUEST OVERRIDES VIA THE ON-LINE FORM: https://www.cics.umass.edu/overrides.