DocumentCode :
3759429
Title :
A Modified K-Means Algorithm Based RBF Neural Network and Its Application in Time Series Modelling
Author :
Yiping Jiao;Yu Shen;Shumin Fei
Author_Institution :
Sch. of Autom., Southeast Univ., Nanjing, China
fYear :
2015
Firstpage :
481
Lastpage :
484
Abstract :
In this paper, a modified K-means based RBFNN is discussed. To improve the performance of RBFNN, an initial cluster centers (ICCs) selection strategy is proposed for K-means algorithm. The algorithm takes breadth preferred subset of samples as ICCs to cover the sample space using greedy strategy. The results shows that the proposed algorithm can improve the performance of RBFNN remarkably in chaotic time series modelling.
Keywords :
"Clustering algorithms","Time series analysis","Machine learning algorithms","Partitioning algorithms","Training","Neural networks","Algorithm design and analysis"
Publisher :
ieee
Conference_Titel :
Distributed Computing and Applications for Business Engineering and Science (DCABES), 2015 14th International Symposium on
Type :
conf
DOI :
10.1109/DCABES.2015.126
Filename :
7429660
Link To Document :
بازگشت