DenMune: Density peak based clustering using mutual nearest neighbors

Many clustering algorithms fail when clusters are of arbitrary shapes, varying densities, or the data classes unbalanced and close to each other, even in two dimensions. A novel algorithm, DenMune is presented meet this challenge. It based on identifying dense regions using mutual nearest neighborhoods size K, where K only parameter required from user, besides obeying neighbor consistency principle. The algorithm stable for a wide range values K. Moreover, it able automatically detect remove noise process as well detecting target clusters. produces robust results various low high-dimensional datasets relative several known state-of-the-art algorithms.