Comparative evaluation of tabu search hyper-heuristic against its low-level meta-heuristic constituents
Hyper-heuristics present a superior form of hybridization of meta-heuristics. Unlike typical meta-heuristic hybridization, which requires low-level integration of two or more metaheuristics, hyper-heuristics offer meta level separation (as domain barrier) of each participating low-level meta-heurist...
| Main Authors: | , |
|---|---|
| Format: | Conference or Workshop Item |
| Language: | English English |
| Published: |
2019
|
| Subjects: | |
| Online Access: | http://umpir.ump.edu.my/id/eprint/25464/ http://umpir.ump.edu.my/id/eprint/25464/1/55.%20Comprative%20evaluation%20of%20tabu%20search%20hyper-heuristic.pdf http://umpir.ump.edu.my/id/eprint/25464/2/55.1%20Comprative%20evaluation%20of%20tabu%20search%20hyper-heuristic.pdf |
| Summary: | Hyper-heuristics present a superior form of hybridization of meta-heuristics. Unlike typical meta-heuristic hybridization, which requires low-level integration of two or more metaheuristics, hyper-heuristics offer meta level separation (as domain barrier) of each participating low-level meta-heuristic and permit adaptive selection between them. Owing to the prospects of improving the generality of its application to general optimization problems, this paper evaluates the performance of a Tabu search based hyper-heuristic (called HHH) against its individual low-level meta-heuristic (LLH) constituents. The results based on its application to t-way test suite generation problem indicate that HHH outperforms all its individual LLH constituents consisting of particle swarm optimization (PSO), global neighbourhood algorithm (GNA), cuckoo search (CS) algorithm and teaching learning-based opthnization algorithm (TLBO). However, there is a time performance penalty as overhead to perform the nmtime adaptive selection of each LLH. |
|---|