Abstract
Achieving an optimal solution for NP-complete problems is a big challenge nowadays. The paper deals with the Traveling Salesman Problem (TSP) one of the most important combinatorial optimization problems in this class. We investigated the Parallel Genetic Algorithm to solve TSP. We proposed a general platform based on Hadoop MapReduce approach for implementing parallel genetic algorithms. Two versions of parallel genetic algorithms (PGA) are implemented, a Parallel Genetic Algorithm with Islands Model (IPGA) and a new model named an Elite Parallel Genetic Algorithm using MapReduce (EPGA) which improve the population diversity of the IPGA. The two PGAs and the sequential version of the algorithm (SGA) were compared in terms of quality of solutions, execution time, speedup and Hadoop overhead. The experimental study revealed that both PGA models outperform the SGA in terms of execution time, solution quality when the problem size is increased. The computational results show that the EPGA model outperforms the IPGA in term of solution quality with almost similar running time for all the considered datasets and clusters. Genetic Algorithms with MapReduce platform provide better performance for solving large-scale problems.
Highlights
Genetic algorithms (GAs) are stochastic search methods that have been successfully applied in many searches, optimization, and machine learning problems [1]
That because we reduced the number of launched jobs during the parallel genetic algorithms (PGA) execution time in order to control the overhead of HDFS accesses, which is limited to the migration phase only
We designed a Hadoop MapReduce platform which is a general framework for implementing parallel genetic algorithms based on two models; the island model and the elite model
Summary
Genetic algorithms (GAs) are stochastic search methods that have been successfully applied in many searches, optimization, and machine learning problems [1]. GAs are used to find approximate solutions in a reasonable time for combinatorial optimization problems. One of the main features of genetic algorithms is that they are inherently parallel. This makes them the most suitable for parallelization [2]. Parallel genetic algorithms (PGAs) can improve GAs to search in a huge solution space and reduce the total execution time. There are three main models of parallel GAs: masterslave model, fine-grained model and coarse-grained called island model
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: International Journal of Advanced Computer Science and Applications
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
