Forschungsthemen
[MA] Systematic Parameter Tuning of Genetic Optimization Approaches
Genetic Optimization approaches have a vast number of unique design choices typically presented in form of parameters. However, a thorough parameter engineering is necessary to enable performance and quality improvements with automatic parameter tuning. The goal of this master thesis is to improve a previously developed genetic optimization approach by carefully tuning its parameters. The research objective is to identify and/or introduce parameters that can significantly influence the quality of the obtained solution, improve performance and increase scalability of the genetic optimization approach.
For this thesis, the following tasks have to be fulfilled:
- Literature analysis covering closely related work.
- Exhaustive analysis of the studied genetic optimization approach, identification of the existing parameters and/or introduction of new design choices.
- Analysis of the aforementioned parameters based with help of a software product line for parameter tuning.
- Evaluation of the optimized genetic optimization approach with a problem of software variant selection and hardware resource allocation.
Betreuer: Dmytro Pukhkaiev