A review on particle swarm optimization algorithms and their applications to data clustering

Abstract

Data clustering is one of the most popular techniques in data mining. It is a method of grouping data into clusters, in which each cluster must have data of great similarity and high dissimilarity with other cluster data. The most popular clustering algorithm K-mean and other classical algorithms suffer from disadvantages of initial centroid selection, local optima, low convergence rate problem etc. Particle Swarm Optimization (PSO) is a population based globalized search algorithm that mimics the capability (cognitive and social behavior) of swarms. PSO produces better results in complicated and multi-peak problems.

Projection-Based Clustering through Self-Organization and Swarm Intelligence

“…The two main types of swarm-based analysis discussed in data science, namely, particle swarm optimization (PSO) and ant colony optimization (ACO) [Martens et al], are distinguished by the type of communication used: PSO agents communicate directly, whereas ACO agents communicate through stigmergy. PSO methods are based on the movement strategies of particles [Kennedy/Eberhart, 1995] and typically used as population-based search algorithms [Rana et al, 2011], whereas ACO methods are applied for sorting tasks [Martens et al, 2011]. In addition to being used to solve discrete optimization problems, PSO has been used as a basis for rulebased classification models, e.g., AntMiner, or as an optimizer within other learning algorithms [Martens et al, 2011], whereas ACO has been used primarily for supervised classification within the data mining community [Martens et al, 2011].…”

“…However, this approach is subject to several of the shortcomings of k-means, which is known to search for spherical clusters [Hennig et al, 2015, p [Ultsch, 2005a]. According to [Rana et al, 2011], the advantages of the clustering process when the PSO approach is used are that it is very fast, simple and easy to understand and implement. "PSO also has very few parameters to adjust [Eberhart et al, 2001] and requires little memory for computation.…”

An Optimized K-Harmonic Means Algorithm Combined with Modified Particle Swarm Optimization and Cuckoo Search Algorithm

“…PSO gives good results and accuracy for single objective optimization, but for multi objective problem, it stuck into local optima. Another PSO problem is its nature to a fast and premature convergence in mid optimum points [38].…”

A Review on Gravitational Search Algorithm and its Applications to Data Clustering & Classification

“…There are many classification techniques that have been developed and applied on the datasets to classify the objects such as decision tree [81,82], Artificial neural network [9,83], A BC [11], PSO [84], Support vector machine [85] etc. Accuracy is the crucial parameter in the classification problems to test the performance of the classificat ion techniques whether the classification techniques properly classified the instance of a datasets or not.…”