Parallel Multi-objective Genetic Algorithm for Classification Rule Mining

摘要

Multi-objective genetic algorithms (MOGAs) are finding increasing popularity as researchers realize their potential for obtaining good solutions to mining problems in large databases. Parallel multi-objective genetic algorithms (pMOGAs) attempts to reduce the processing time needed for computing the fitness functions and to reach an acceptable solution. We propose two different master slave models of pMOGA. Our proposed models exploit both data parallelism by distributing the data being mined across various processors, and control parallelism by distributing the population of individuals across ail available processors. These models are implemented through a cluster computing environment and we measure the speed up of pMOGA over its sequential counterpart.