Kernel-based adaptive K-Medoid clustering

Aiming at the weakness of K-Medoid algorithm that it is unable effectively to cluster big data set and high-dimension data, kernel-based learningmethod is introduced to the K-Medoid clustering algorithm, kernel-based adaptiveK-Medoid algorithm is proposed, the algorithm firstly maps the data from their input space to a high dimensional feature space by kernel function, then performs K-Medoid clustering in the high dimensional space. In data cluster process, data can join the optimum cluster, and the cluster result has no relation with how to select the originalk centers. Thenew algorithm can cluster largedata set and high-dimensiondata. The results ofexperiments show that kernel-based adaptive K-Medoid algorithm has better precision ratio than K-Medoid algorithm.