A Fuzzy Clustering Algorithm Based on Weighted Index and Optimization of Clustering Number
In view of the discretization of continuous attributes of civil aviation radar intelligence data, this paper proposes a fuzzy partition algorithm of continuous attributes based on weighted index and optimization of clustering number, and its automatic det
- PDF / 495,407 Bytes
- 11 Pages / 439.37 x 666.142 pts Page_size
- 42 Downloads / 217 Views
Abstract In view of the discretization of continuous attributes of civil aviation radar intelligence data, this paper proposes a fuzzy partition algorithm of continuous attributes based on weighted index and optimization of clustering number, and its automatic determination of optimal weighted index m* and optimal clustering number c* overcomes the shortcomings of current attribute fuzzy methods of manual determination of classification number and no consideration of geometry data. The experimental results verify the validity and feasibility of fuzzy attribute discretization of civil aviation radar intelligence data characteristics. Keywords Continuous attributes clustering algorithm
Weighted index Clustering number Fuzzy
1 Introduction Fuzzy clustering is an important branch of unsupervised pattern recognition and has been widely applied in the field of image processing, pattern recognition, time series prediction, and parameter estimation, and in practice, fuzzy clustering means (FCM) has been the most widely used fuzzy clustering algorithm. FCM algorithm is an unsupervised classification algorithm, and the weighted index m* and clustering number c*, its two important parameters, should be determined in advance. If the parameters are inappropriately selected, the classification results may not match with the real structure of data. According to some literatures, [1] ‘‘parameter m controls the sharing degree between the fuzzy classes.’’ Currently, implementation of FCM algorithm is usually based on experience or experiment and artificial selection method to determine the initial parameters of W. Wang (&) Q. Li Air Force Early Warning Academy, Wuhan 430000, Hubei, China e-mail: [email protected]
Z. Wen and T. Li (eds.), Foundations of Intelligent Systems, Advances in Intelligent Systems and Computing 277, DOI: 10.1007/978-3-642-54924-3_33, Springer-Verlag Berlin Heidelberg 2014
349
350
W. Wang and Q. Li
the algorithms for clustering; however, how to select the optimum m and optimal c has not been solved effectively. In view of the above problems, according to the fuzzy decision theory and fuzzy clustering validity index Vkwon of data geometric structure, the weighted index m and c are discussed, and the methods to realize optimum parameters and a fuzzy clustering algorithm are based on weighted index and clustering number optimization. As for application of FCM clustering algorithm, the weighted index and clustering number should be specified manually.
2 Description of FCM Clustering Algorithm FCM clustering algorithm obtains membership of each sample site to the sample center through optimizing objective function Jm, so as to determine the membership of sample site. Objective function-based fuzzy clustering method is put forward firstly by Ruspini [2], but the real effective method is proposed by Dunn [3] in 1974. Hard c-means (HCM) clustering algorithm is extended to the fuzzy case by Dunn. In that same year, Bezdek generalizes Dunn’s method and establishes the FCM clustering theory. Through optim
Data Loading...
