DocumentCode :
2971795
Title :
Nearest Neighbor-Clustering Algorithm Based on Hierarchical Optimization Strategy
Author :
Jie Wang ; Guoqiang Jiang
Author_Institution :
Coll. of Electr. Eng., Zhengzhou Univ., Zhengzhou
fYear :
2008
fDate :
2-3 Aug. 2008
Firstpage :
233
Lastpage :
236
Abstract :
In order to overcome the shortcoming of nearest neighbor-clustering algorithm in the cluster center determined, the cluster width of the acquisition, and the hidden nodes learning. A FCM strategy is being proposed to determine the cluster center, introducing the target function and the LMS method to make the cluster width adjusted adaptively, and a pruning strategy is adopted to cut the redundant hidden nodes. The simulation results in the nearest neighbor-clustering based on hierarchical optimization strategy show that the algorithms are greatly improved in the learning accuracy and speed.
Keywords :
least mean squares methods; optimisation; radial basis function networks; LMS method; hierarchical optimization strategy; nearest neighbor-clustering algorithm; pruning strategy; radial basis function; redundant hidden nodes; Algorithm design and analysis; Clustering algorithms; Design optimization; Educational institutions; Function approximation; Intelligent transportation systems; Least squares approximation; Neural networks; Power electronics; Signal processing algorithms; FCM; LMS; nearest neighbor-clustering algorithm; pruning strategy;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Power Electronics and Intelligent Transportation System, 2008. PEITS '08. Workshop on
Conference_Location :
Guangzhou
Print_ISBN :
978-0-7695-3342-1
Type :
conf
DOI :
10.1109/PEITS.2008.55
Filename :
4634850
Link To Document :
بازگشت