Abstract
Aim: The main aim of this paper is to improve the clustering efficiency in standard medical and non-medical datasets. This is achieved using a new optimization technique named as IQPSO-FCM which is a combination of Fuzzy C-Means and Improved Quantum Particle Swarm Optimization that use Stern series to obtain new global best positions of particles. Methodology: Fuzzy clustering is an important method for clustering, which groups the data elements based on their degree of membership and the elements can belong to more than one cluster signifying that a person can have more than one disease. Fuzzy C-Means works well for noise free data sets but sometimes fails when original data is corrupted with noise. The proposed IQPSO-FCM clusters the data elements by finding global best positions using IQPSO. These gbest values are considered as cluster centers for Fuzzy CMeans. Fuzzy C-Means in turn groups the elements based on Euclidean distance. The experiments were done on five benchmark datasets, two medical datasets namely liver disorder and breast cancer and three general datasets namely iris, glass and wine from UCI machine learning repository. Results: The experimental results of proposed algorithm are compared with standard algorithms using intra and inter clustering distances and time elapsed using MATLAB. The execution time of FCM and PCA-FCM is lesser than proposed method but the accuracy of outcome is less. The results on cluster size of 2, 3, 4 and 5 showed that the proposed
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call