Open Access. Powered by Scholars. Published by Universities.®
![Digital Commons Network](http://assets.bepress.com/20200205/img/dcn/DCsunburst.png)
Operations Research, Systems Engineering and Industrial Engineering Commons™
Open Access. Powered by Scholars. Published by Universities.®
- Publication
- Publication Type
Articles 1 - 2 of 2
Full-Text Articles in Operations Research, Systems Engineering and Industrial Engineering
Generic Instance-Specific Automated Parameter Tuning Framework, Linda Lindawati
Generic Instance-Specific Automated Parameter Tuning Framework, Linda Lindawati
Dissertations and Theses Collection (Open Access)
Meta-heuristic algorithms play an important role in solving combinatorial optimization problems (COP) in many practical applications. The caveat is that the performance of these meta-heuristic algorithms is highly dependent on their parameter configuration which controls the algorithm behaviour. Selecting the best parameter configuration is often a difficult, tedious and unsatisfying task. This thesis studies the problem of automating the selection of good parameter configurations. Existing approaches to address the challenges of parameter configuration can be classified into one-size-fits-all and instance-specific approaches. One-size-fits-all approaches focus on finding a single best parameter configuration for a set of problem instances, while instance-specific approaches …
Automated Parameter Tuning Framework For Heterogeneous And Large Instances: Case Study In Quadratic Assignment Problem, Linda Lindawati, Zhi Yuan, Hoong Chuin Lau, Feida Zhu
Automated Parameter Tuning Framework For Heterogeneous And Large Instances: Case Study In Quadratic Assignment Problem, Linda Lindawati, Zhi Yuan, Hoong Chuin Lau, Feida Zhu
Research Collection School Of Computing and Information Systems
This paper is concerned with automated tuning of parameters of algorithms to handle heterogeneous and large instances. We propose an automated parameter tuning framework with the capability to provide instance-specific parameter configurations. We report preliminary results on the Quadratic Assignment Problem (QAP) and show that our framework provides a significant improvement on solutions qualities with much smaller tuning computational time.