Mining Fuzzy Association Rules Based on Parallel Particle Swarm Optimization Algorithm

Abstract

The association rule extraction process often involves a large number of candidate item sets and multiple read operations on data sets. With the emergence of massive data, the sequential association rule extraction algorithm also suffers from large I/O overhead and insufficient memory. This paper presents a new multi-swarm parallel multi-mutation particle swarm optimization algorithm (MsP-MmPSO) to search several groups in parallel. Experimental results show that the MsP-MmPSO algorithm has an advantage in terms of execution time over traditional particle swarm optimization, especially when the amount or dimensions of the data increase. Experiments also verify that a good task allocation method can reduce the execution time of the parallel algorithm.