University of Twente Student Theses


Dynamic Parameter Tuning Method With Presets For Metaheuristics

Wei, Kaiyu (2022) Dynamic Parameter Tuning Method With Presets For Metaheuristics.

[img] PDF
Abstract:In this paper, we design an online learning method to select the proper combination of parameters to give the running metaheuristic a good performance without pilot tests. The method runs simultaneously with the metaheuristic and learns about the attributes of different parameters and their values, named as Dynamic parameter value tuning method (DPTP). The advantage of this method is that it needs no pilot test before the metaheuristic is run for selecting proper parameter values for it. This cuts down on running time for users and gives a good selection of parameter values. Existing parameter-control methods mostly handle the situation where there is only one parameter, while DPTP can manage multiple parameters at the same time. We provide “presets”, which are alternative combinations of parameter values, for DPTP so that it can dynamically select different value settings and apply the selected values to the metaheuristic.
Item Type:Essay (Master)
Dassault Systèmes, 's-Hertogenbosch, Netherlands
Faculty:BMS: Behavioural, Management and Social Sciences
Subject:31 mathematics, 54 computer science
Programme:Industrial Engineering and Management MSc (60029)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page