Sustainable Implementation of Neuro-Fuzzy Systems Based Clustering Using Hybrid Learning Vector Quantiza-tion

Abstract

In current years, key inventions in Neuro Inference System research have transformed the machine learning background from an engineering perception. Neuro Inference System is hominid ready information handling systems that are grown up extensively in the last thirty years. The proposed work uses the Hybridized Distance function along with different distance functions viz. Euclidean, Canberra and Chebyshev function in Learning Vector Quantization (LVQ) algorithm and observes the clustering results in Breast Cancer dataset.  In this work, we use the Supervised Learning Vector Quantization technique on hybrid distance function which work in both supervised as well as unsupervised modes and it is a heuristic search technique based on the patents available. The data sets have been taken form Medical Science for providing learning and examining. It increases the effective finding of diseases by using various data classification methods. Thus, it modifies the Neuro Inference System algorithm to improve its performance. The experimental work is simulated with simulation software MATLAB 7. After a comparative analysis of the results, it observed that the Hybrid LVQ is more accurate as compared to Canberra and conventional Euclidean Distance. The proposed Hybrid LVQ performance is enhanced up to 90.6% from 87.1% of Canberra Distance. It shows that Canberra gives the minimum distance between points from centroid as compared to Euclidean and Chebyshev Distance Function. The proposed Distance Function method gives the minimum distance of points from centroid as compared to Canberra Distance Function.