This seminar will be implemented as a block seminar, likely online.

Location and schedule:

• first meeting (introduction and paper assignment): TBA, at the start of the term, at ZIB (Takustr. 7), Seminar room 2006 (ground floor)
• second meeting (short talks): TBA in the middle of the term, probably at ZIB, one hour
• final meeting (seminar talks): TBA on one or two days at the end of the semester

Students should have some background in mathematical optimization including basic graph theory and basic linear programming.

In the middle of the term, you are supposed to give a short, introductory talk (at most 5 minutes) on your topic.

The talk must include a literature search, we must encompass at least a forward and a backward analysis.

To obtain the credit points, you are also required to hand in a short summary of your talk (use LaTeX, 5-8 pages). The summary should be sent by e-mail to your advisor. The summary will be graded and then handed back to you. We hope that this feedback will enable you to give a better presentation.

The seminar itself will take place on one or two days in the last weeks of the semester. Talks should be prepared for 45 minutes, so that a duration of 60 minutes including questions is not exceeded. Participation in the *entire* seminar (not only on the day of your own talk) and timely submission of a summary is a requirement for passing.

Your final grade will be composed of 60% and 40% from the evaluation of your talk and paper, respectively.

The combination of methods from combinatorial optimization and machine learning offers a potential to deal with fluctuations in suppy and demand, to master operational problems such as delays etc. The idea is that smart strategies that involve data based anticipation of future developments will lead to better decisions.This seminar will study such methods on the interface between optimization and machine learning by means of a selection of recent research papers. We will focus on applications in air and train freight management with particularly large potentials.