Scheduling

Motivation

Scheduling is a major research field that is studied from a practical and theoretical perspective in computer science, mathematical optimization and operations research. Applications range from traditional production scheduling and project planning to the newly arising resource management tasks in the area of internet technology such as distributed cloud service networks and the allocation of virtual machines to physical servers. Algorithms for scheduling problems have been one of the richest areas of algorithmic research, spanning nearly 70 years of work. Throughout, research has been prompted by the fact that in most settings, these computational problems are quite challenging, and new approaches and frameworks are continually being added to help tackle a broadening portfolio of scheduling problems.

At this Dagstuhl Seminar, we focus on the emerging models for beyond-worst case algorithm design, in particular, recent approaches that incorporate learning. Several models for the integration of learning into algorithm design have been proposed and have already demonstrated advances in the state-of-art for various scheduling applications. This seminar will focus on three established themes:

1. Scheduling with error-prone learned predictions
2. Data-driven algorithm design
3. Stochastic and bayesian learning in scheduling

The seminar aims to bring together researchers working on distinct areas to encourage cross-fertilization among different research directions. As the field of learning is very broad, we methodologically focus on the theoretical design of algorithms with provable performance guarantees.