A Nearest Neighbor Similarity based Method for Robust Clustering

• Main Authors: Chun-You Cho, 卓俊佑
• Other Authors: Jin-Shiuh Taur
• Format: Others
• Language: zh-TW
• Published: 2007
• Online Access: http://ndltd.ncl.edu.tw/handle/90474591749017961626

碩士 === 中興大學 === 電機工程學系所 === 95 === This research first discusses several problems that often occur in conventional clustering analysis. The capability of solving these problems can be used as the performance index of clustering methods. Then basic theories of several conventional clustering methods are reviewed and the survey of advantages and disadvantages is included. Based on these discussions, we introduce a nearest neighbor similarity-based robust clustering method (NN-SCM) by discovering the fundamental structure of the data set. Finally, several different data sets are used to compare the performances of the proposed approach and the traditional similarity-based clustering method. Simulation results indicate that the proposed approach has better clustering performance with less computation time while avoiding most of the conventional clustering problems.