Proposed Enhancement in the Performance of the Spectral Clustering

Abstract- Spectral clustering has become one of the most hotspots in clustering over the past few years. It has been proved to be effective by the many researchers and its performance outperforms many other clustering techniques. Much work has been done to enhance the performance of the spectral clustering but still many of them are quite computationally expensive. In this work, focus will be on reducing the computation time and also on the accuracy in order to improve the quality of clusters. The base paper used the Gaussian kernel functions to construct the similarity matrix and in this work some other light weight functions will be used instead of the Gaussian kernel functions and it is expected to be effective in achieving the expected goals. Keywords- Spectral clustering, computation time, quality of clusters. I