The preservation of favored building blocks in the struggle for fitness: The puzzle algorithm

Research output: Contribution to journalArticlepeer-review

18 Scopus citations

Abstract

The shortest common superstring (SCS) problem, own to be NP-complete, seeks the shortest string that contains all strings from a given set. In this paper, we present a novel coevolutionary algorithm-the Puzzle Algorithm-where a population of building blocks coevolves alongside a population of solutions. We show experimentally that our novel algorithm outperforms a standard genetic algorithm (GA) and a benchmark greedy algorithm on instances of the SCS problem inspired by deoxyribonucleic acid (DNA) sequencing. We next compare our previously presented cooperative coevolutionary algorithm with the Co-Puzzle Algorithm-the puzzle algorithm coupled with cooperative coevolution-showing that the latter proves to be top gun. Finally, we discuss the benefits of using our puzzle approach in the general field of evolutionary algorithms.

Original languageEnglish
Pages (from-to)443-455
Number of pages13
JournalIEEE Transactions on Evolutionary Computation
Volume8
Issue number5
DOIs
StatePublished - 1 Oct 2004

Fingerprint

Dive into the research topics of 'The preservation of favored building blocks in the struggle for fitness: The puzzle algorithm'. Together they form a unique fingerprint.

Cite this