A hybrid feature selection and classification strategy was proposed based on the simulated annealing genetic algonthrn and multiple instance learning （MIL）. The band selection method was proposed from subspace decom...A hybrid feature selection and classification strategy was proposed based on the simulated annealing genetic algonthrn and multiple instance learning （MIL）. The band selection method was proposed from subspace decomposition, which combines the simulated annealing algorithm with the genetic algorithm in choosing different cross-over and mutation probabilities, as well as mutation individuals. Then MIL was combined with image segmentation, clustering and support vector machine algorithms to classify hyperspectral image. The experimental results show that this proposed method can get high classification accuracy of 93.13% at small training samples and the weaknesses of the conventional methods are overcome.展开更多
This paper mainly studies case representation based on fuzzy technology, the comparing framework of case properties and the method of similarity assessment. Firstly, this paper proposes the method of case representati...This paper mainly studies case representation based on fuzzy technology, the comparing framework of case properties and the method of similarity assessment. Firstly, this paper proposes the method of case representation based on fuzzy technology. Secondly, it discusses the comparing framework of case fuzzy properties. Thirdly, it presents the case representation framework composed of structure data and related properties and b＇ased on which it studies the method and step of gaining case similarity assessment through adjusting similarity value of a property. This paper proposes a similarity assessment method in comparing stage of CBR using fuzzy technology. This method not only simples the computing complexity of CBR in comparing stage but also resolves the difficulties of similarity assessment on different levels.展开更多
This paper proposes a moving ratio test, which is a modified version of ratio test by involving a bandwidth parameter, to detect multiple changes in persistence, and to determine the direction of switch. The asymptoti...This paper proposes a moving ratio test, which is a modified version of ratio test by involving a bandwidth parameter, to detect multiple changes in persistence, and to determine the direction of switch. The asymptotic distribution as well as its consistency are established. In order to estimate the change points, the authors propose a two step estimate approach. Simulations illustrate the performance of the test and estimate procedures in finite samples. An application to the analysis of Sweden/US foreign exchange rate demonstrates the validity of the procedure in practice.展开更多
文摘A hybrid feature selection and classification strategy was proposed based on the simulated annealing genetic algonthrn and multiple instance learning （MIL）. The band selection method was proposed from subspace decomposition, which combines the simulated annealing algorithm with the genetic algorithm in choosing different cross-over and mutation probabilities, as well as mutation individuals. Then MIL was combined with image segmentation, clustering and support vector machine algorithms to classify hyperspectral image. The experimental results show that this proposed method can get high classification accuracy of 93.13% at small training samples and the weaknesses of the conventional methods are overcome.
文摘This paper mainly studies case representation based on fuzzy technology, the comparing framework of case properties and the method of similarity assessment. Firstly, this paper proposes the method of case representation based on fuzzy technology. Secondly, it discusses the comparing framework of case fuzzy properties. Thirdly, it presents the case representation framework composed of structure data and related properties and b＇ased on which it studies the method and step of gaining case similarity assessment through adjusting similarity value of a property. This paper proposes a similarity assessment method in comparing stage of CBR using fuzzy technology. This method not only simples the computing complexity of CBR in comparing stage but also resolves the difficulties of similarity assessment on different levels.
基金supported by the National Natural Science Foundation of China under Grant Nos.61065009 and 60972150
文摘This paper proposes a moving ratio test, which is a modified version of ratio test by involving a bandwidth parameter, to detect multiple changes in persistence, and to determine the direction of switch. The asymptotic distribution as well as its consistency are established. In order to estimate the change points, the authors propose a two step estimate approach. Simulations illustrate the performance of the test and estimate procedures in finite samples. An application to the analysis of Sweden/US foreign exchange rate demonstrates the validity of the procedure in practice.
