Forschungsthemen
[MA] Capturing Variability of Dynamic Multi-model Surrogate Optimization
Dynamic multi-model surrogate optimization is a recent approach for expensive black-box optimization adhering to the principles of software product line engineering. However, capturing its variability with existing feature modelling approaches is challenging. Although the combined set of variability management mechanisms, provided by the state-of-the-art approaches, is extensive; no approach provides a full set of mechanisms to perform dynamic multi-model surrogate optimization. The goal of this thesis is to extend the capabilities of the existing variability management approaches. The research objective is to identify the lacking variability management mechanisms in the existing approaches and to handle them within a single approach able to capture the variability model of dynamic multi-model surrogate optimization. For this thesis the following tasks have to be fulfilled: * Literature analysis covering closely related work. * Identification of the missing variablity management mechanisms in state-of-the-art approaches. * Development of the approach capturing the variability model of dynamic multi-model surrogate optimization. * Implementation of the developed approach. * Evaluation of the developed approach.
Betreuer: Dmytro Pukhkaiev