Clustering Via Local Regression

This paper deals with the local learning approach for clustering, which is based on the idea that in a good clustering, the cluster label of each data point can be well predicted based on its neighbors and their cluster labels. We propose a novel local learning based clustering algorithm using kernel regression as the local label predictor. Although sum of absolute error is used instead of sum of squared error, we still obtain an algorithm that clusters the data by exploiting the eigen-structure of a sparse matrix. Experimental results on many data sets demonstrate the effectiveness and potential of the proposed method.