General Information

Course instructors: Andreas S. Schulz, Felix Buld

Many discrete optimization problems require decisions to be made without full knowledge of future inputs. Online algorithms address this challenge, traditionally evaluated by comparing their performance to optimal solutions of underlying offline problems. Recently, the study of learning-augmented algorithms has become a very vibrant research area. These algorithms aim to enhance performance by utilizing predictions, while still ensuring robust worst-case guarantees. In this seminar, we will explore interesting classical and recent results at the transition from offline to online optimization, with applications including scheduling, optimal stopping, mechanism design, and facility location.

Some helpful advice for your presentation can be found here.