Determining the number of components is a crucial issue in a mixture model. A moment-based criterion is considered to estimate the number of components arising from a normal mixture model. This criterion is derived fr...Determining the number of components is a crucial issue in a mixture model. A moment-based criterion is considered to estimate the number of components arising from a normal mixture model. This criterion is derived from an omnibus statistic involving the skewness and kurtosis of each component. The proposed criterion additionally provides a measurement for the model fit in an absolute sense. The performances of our criterion are satisfactory compared with other classical criteria through Monte-Carlo experiments.展开更多
A new fuzzy support vector machine algorithm with dual membership values based on spectral clustering method is pro- posed to overcome the shortcoming of the normal support vector machine algorithm, which divides the ...A new fuzzy support vector machine algorithm with dual membership values based on spectral clustering method is pro- posed to overcome the shortcoming of the normal support vector machine algorithm, which divides the training datasets into two absolutely exclusive classes in the binary classification, ignoring the possibility of ＂overlapping＂ region between the two training classes. The proposed method handles sample ＂overlap＂ effi- ciently with spectral clustering, overcoming the disadvantages of over-fitting well, and improving the data mining efficiency greatly. Simulation provides clear evidences to the new method.展开更多
基金supported by the National Natural Sciences Foundation of China(7137102271401193+2 种基金71671193)the Program for Innovation Research in Central University of Finance and Economicsthe Innovation Foundation of BUAA for Ph.D.Graduates
文摘Determining the number of components is a crucial issue in a mixture model. A moment-based criterion is considered to estimate the number of components arising from a normal mixture model. This criterion is derived from an omnibus statistic involving the skewness and kurtosis of each component. The proposed criterion additionally provides a measurement for the model fit in an absolute sense. The performances of our criterion are satisfactory compared with other classical criteria through Monte-Carlo experiments.
基金supported by the National Natural Science Foundation of China (7083100170821061)
文摘A new fuzzy support vector machine algorithm with dual membership values based on spectral clustering method is pro- posed to overcome the shortcoming of the normal support vector machine algorithm, which divides the training datasets into two absolutely exclusive classes in the binary classification, ignoring the possibility of ＂overlapping＂ region between the two training classes. The proposed method handles sample ＂overlap＂ effi- ciently with spectral clustering, overcoming the disadvantages of over-fitting well, and improving the data mining efficiency greatly. Simulation provides clear evidences to the new method.
