Swarm intelligence based intuitionistic fuzzy c means clustering algorithms

Abstract

Clustering is the process of grouping data objects based on their characteristics. Clustering is an unsupervised technique wherein there is no initial knowledge about the dataset is available. Fuzzy clustering allows an object to be a part of more than one cluster based on the membership values. Intuitionistic fuzzy clustering relies on the fact that the uncertainty in data can be well represented as hesitancy or indeterminancy that indicates the ignorance of user about the data. The issues in clustering are sensitivity to selection of initial centroids, need to specify number of clusters while starting the execution of clustering algorithm, tendency towards local optimal solutions and so on. To address these research issues, intuitionistic fuzzy C-means clustering algorithms are hybridized with the three famous swarm intelligence techniques named particle swarm optimization, cuckoo search algorithm and crow search algorithm. These hybrid algorithms are tested over numerical and image data. Particle swarm optimization plays a prominent role in the selection of optimal centroids. The initial set of centroids is considered as the particles and they move with a particular velocity towards the solution. The candidate solutions are identified and the best feasible among them is chosen. This set of centroids is then given as the initial seeds to the intuitionistic fuzzy clustering algorithm. The problem with particle swarm optimization is that it exhibits slow convergence with real-time problems. Further, the computational cost is directly proportional to the number of clusters and volume of data. newline