Title :
Entropy regularized fuzzy C-lines for data with tolerance
Author :
Kanzawa, Yuchi ; Endo, Yasunori ; Miyamoto, Sadaaki
Author_Institution :
Dept. of Risk Eng., Univ. of Tsukuba, Tsukuba, Japan
Abstract :
This paper presents a new clustering algorithm, which is based on entropy regularized fuzzy c-lines, can treat data with some errors. First, the tolerance is formulated and introduce into optimization problem of clustering. Next, the problem is solved using Karush-Kuhn-Tucker conditions. Last, the algorithm is constructed based on the results of solving the problem. Some numerical examples for the proposed method are shown.
Keywords :
fuzzy set theory; optimisation; pattern clustering; Karush-Kuhn-Tucker condition; clustering algorithm; entropy regularized fuzzy C-lines; optimization problem; pattern clustering; Clustering algorithms; Computer errors; Design engineering; Entropy; Manifolds; Principal component analysis; Prototypes; Systems engineering and theory; Virtual colonoscopy;
Conference_Titel :
Fuzzy Systems, 2009. FUZZ-IEEE 2009. IEEE International Conference on
Conference_Location :
Jeju Island
Print_ISBN :
978-1-4244-3596-8
Electronic_ISBN :
1098-7584
DOI :
10.1109/FUZZY.2009.5277176
