Analysis of evolutionary computing performance via mapreduce parallel processing architecture / Ahmad Firdaus Ahmad Fadzil

Ahmad, Ahmad Firdaus (2014) Analysis of evolutionary computing performance via mapreduce parallel processing architecture / Ahmad Firdaus Ahmad Fadzil. Masters thesis, Universiti Teknologi MARA.

[img]
Preview
Text
TM_AHMAD FIRDAUS BIN AHMAD FADZIL CS 14_5 1.pdf

Download (192kB) | Preview

Abstract

Evolutionary computation (EC) is a method that is ubiquitously used to solve complex computation. Examples of EC such as Genetic Algorithm (GA) and PSO (Particle Swarm Optimization) are prevalent due to their efficiency and effectiveness. Despite these advantages, EC suffers from long execution time due to its parallel nature. Therefore, this research explores the prospect of speeding up the EC algorithms specifically GA and PSO via MapReduce (MR) parallel processing framework. MR is an emerging parallel processing framework that hides the complex parallelization processes by employing the functional abstraction of "map and reduce" The Performance of the parallelized GA via MR and PSO via MR are evaluated using an analogous case study to find out the speedup and efficiency in order to measure the scalability of both proposed algorithms. Comparisons between GA via MR and PSO via MR are also established in order to find which EC algorithm scales better via MR parallel processing framework. From the results and analysis obtained from this research, it is established that both GA and PSO can be efficiently parallelized and shows good scalability via MR parallel processing framework. The Performance comparison between GA via MR and PSO via MR also shows that both algorithms are comparable in terms of speedup and efficiency

Item Type: Thesis (Masters)
Subjects: Q Science > QA Mathematics > Instruments and machines > Electronic computers.Computer science
Q Science > QA Mathematics > Instruments and machines > Electronic computers.Computer science
Divisions: Faculty of Computer and Mathematical Sciences
Depositing User: Staf Pendigitan 2
Date Deposited: 23 Apr 2015 02:54
Last Modified: 14 May 2016 07:20
URI: http://ir.uitm.edu.my/id/eprint/11938

Actions (login required)

View Item View Item

Downloads

Downloads per month over past year