Title :
A fuzzy and locally sensitive method for cluster analysis
Author :
Thome, Antonio Carlos Gay
Author_Institution :
Electr. Eng. Dept., Mil. Inst. of Technol., Rio de Janiero, Brazil
Abstract :
Cluster analysis has been playing an important role in pattern recognition, image processing and time series analysis. The majority of the existing clustering algorithms depend on initial parameters and assumptions about the underlying data structure. A fuzzy method of mode separation is proposed. The method addresses the task of multi-modal partitioning through a sequence of locally sensitive searches guided by a stochastic gradient ascent procedure, and addresses the cluster validity problem through a global partition performance criterion. The algorithm is computationally efficient and provides good results when tested with a number of simulated and real data sets
Keywords :
fuzzy set theory; pattern recognition; search problems; stochastic processes; time series; cluster analysis; cluster validity problem; clustering algorithms; data structure; fuzzy method; global partition performance criterion; image processing; locally sensitive method; locally sensitive search; mode separation; multimodal partitioning; pattern recognition; stochastic gradient ascent procedure; time series analysis; Clustering algorithms; Data structures; Image analysis; Image processing; Partitioning algorithms; Pattern analysis; Pattern recognition; Stochastic processes; Testing; Time series analysis;
Conference_Titel :
Intelligent Information Systems, 1995. ANZIIS-95. Proceedings of the Third Australian and New Zealand Conference on
Conference_Location :
Perth, WA
Print_ISBN :
0-86422-430-3
DOI :
10.1109/ANZIIS.1995.705755
