Improvement of the fast exact pairwise-nearest-neighbor algorithm

摘要

Pairwise-nearest-neighbor (PNN) is an effective method of data clustering, which can always generate good clustering results, but with high computational complexity. Fast exact PNN (FPNN) algorithm proposed by Franti et al. is an effective method to speed up PNN and generates the same Clustering results as those generated by PNN. In this paper, We present a novel method to improve the FPNN algorithm. Our algorithm uses the property that the cluster distance increases as the cluster merge process proceeds and adopts a fast search algorithm to reject impossible candidate clusters. Experimental results show that Our proposed method can effectively reduce the number of distance calculations and computation time of FPNN algorithm. Compared with FPNN, Our proposed approach can reduce the computation time and number of distance calculations by a factor of 24.8 and 146.4, respectively, for the data set from three real images. It is noted that our method generates the same clustering results as those produced by PNN and FPNN.