Soft Computing

, 13:871

A case study of memetic algorithms for constraint optimization


DOI: 10.1007/s00500-008-0354-4

Cite this article as:
Özcan, E. & Başaran, C. Soft Comput (2009) 13: 871. doi:10.1007/s00500-008-0354-4


There is a variety of knapsack problems in the literature. Multidimensional 0–1 knapsack problem (MKP) is an NP-hard combinatorial optimization problem having many application areas. Many approaches have been proposed for solving this problem. In this paper, an empirical investigation of memetic algorithms (MAs) that hybridize genetic algorithms (GAs) with hill climbing for solving MKPs is provided. Two distinct sets of experiments are performed. During the initial experiments, MA parameters are tuned. GA and four MAs each using a different hill climbing method based on the same configuration are evaluated. In the second set of experiments, a self-adaptive (co-evolving) multimeme memetic algorithm (MMA) is compared to the best MA from the parameter tuning experiments. MMA utilizes the evolutionary process as a learning mechanism for choosing the appropriate hill climbing method to improve a candidate solution at a given time. Two well-known MKP benchmarks are used during the experiments.


Evolutionary algorithms Self-generation Knapsack problem Local search Adaptation 

Copyright information

© Springer-Verlag 2008

Authors and Affiliations

  1. 1.Department of Computer EngineeringYeditepe UniversityKadıköy, IstanbulTurkey

Personalised recommendations