Automated Algorithm Selection

For the solution of an optimization problem, there often is a diverse portfolio of solution algorithms. Classically, in an application scenario, the specific algorithm has to be selected from this portfolio or even configured by an expert.

Automated Algorithm Selection (AAS) aims at automatically selecting a solution algorithm from this portfolio. This is usually done by a feature-based problem characterization of the problem instances and machine learning. AAS thus enables novices to effectively apply modern and well-performing solution heuristics without in-depth knowledge of their methodology.