KNN-SC: Novel Spectral Clustering Algorithm Using k-Nearest Neighbors

Spectral clustering is a well-known graph-theoretic algorithm. Although spectral has several desirable advantages (such as the capability of discovering non-convex clusters and applicability to any data type), it often leads incorrect results because high sensitivity noise points. In this study, we propose robust algorithm known KNN-SC that can discover exact by decreasing influence To achieve goal, present novel approach filters out potential points estimating density difference between using $k$ -nearest neighbors. addition, introduce method for generating similarity graph in which various densities are effectively represented expanding nearest neighbor graph. Experimental on synthetic real-world datasets demonstrate achieves significant performance improvement over many state-of-the-art algorithms.