The validity measurement of fuzzy clustering is a key problem. If clustering is formed, it needs a kind of machine to verify its validity. To make mining more accountable, comprehensible and with a usable spatial patt...The validity measurement of fuzzy clustering is a key problem. If clustering is formed, it needs a kind of machine to verify its validity. To make mining more accountable, comprehensible and with a usable spatial pattern, it is necessary to first detect whether the data set has a clustered structure or not before clustering. This paper discusses a detection method for clustered patterns and a fuzzy clustering algorithm, and studies the validity function of the result produced by fuzzy clustering based on two aspects, which reflect the un-certainty of classification during fuzzy partition and spatial location features of spatial data, and proposes a new validity function of fuzzy clustering for spatial data. The experimental result indicates that the new validity function can accurately measure the validity of the results of fuzzy clustering. Especially, for the result of fuzzy clustering of spatial data, it is robust and its classification result is better when compared to other indices.展开更多
Analysis of ocean fronts＇ uncertainties indicates that they result from indiscemibility of their spatial position and fuzziness of their intensity. In view of this, a flow hierarchy for uncertainty representation of ...Analysis of ocean fronts＇ uncertainties indicates that they result from indiscemibility of their spatial position and fuzziness of their intensity. In view of this, a flow hierarchy for uncertainty representation of ocean fronts is proposed on the basis of fuzzy-rough set theory. Firstly, raster scanning and blurring are carried out on an ocean front, and the upper and lower approximate sets, the indiscernible relation in fuzzy-rough theories and related operators in fuzzy set theories are adopted to represent its uncertainties, then they are classified into three sets： with members one hundred percent belonging to the ocean front, belonging to the ocean front＇s edge and definitely not belonging to the ocean front. Finally, the approximate precision and roughness degree are utilized to evaluate the ocean front＇s degree of uncertainties and the precision of the representation. It has been proven that the method is not only capable of representing ocean fronts＇ uncertainties, but also provides a new theory and method for uncertainty representation of other oceanic phenomena.展开更多
文摘The validity measurement of fuzzy clustering is a key problem. If clustering is formed, it needs a kind of machine to verify its validity. To make mining more accountable, comprehensible and with a usable spatial pattern, it is necessary to first detect whether the data set has a clustered structure or not before clustering. This paper discusses a detection method for clustered patterns and a fuzzy clustering algorithm, and studies the validity function of the result produced by fuzzy clustering based on two aspects, which reflect the un-certainty of classification during fuzzy partition and spatial location features of spatial data, and proposes a new validity function of fuzzy clustering for spatial data. The experimental result indicates that the new validity function can accurately measure the validity of the results of fuzzy clustering. Especially, for the result of fuzzy clustering of spatial data, it is robust and its classification result is better when compared to other indices.
基金The research was partially funded by the Project 40571129 supported by the National Natural Science Foundation of ChinaInnovative Program(No.kzcx2-yw-304-1)supported by the Chinese Academy of Sciences.
文摘Analysis of ocean fronts＇ uncertainties indicates that they result from indiscemibility of their spatial position and fuzziness of their intensity. In view of this, a flow hierarchy for uncertainty representation of ocean fronts is proposed on the basis of fuzzy-rough set theory. Firstly, raster scanning and blurring are carried out on an ocean front, and the upper and lower approximate sets, the indiscernible relation in fuzzy-rough theories and related operators in fuzzy set theories are adopted to represent its uncertainties, then they are classified into three sets： with members one hundred percent belonging to the ocean front, belonging to the ocean front＇s edge and definitely not belonging to the ocean front. Finally, the approximate precision and roughness degree are utilized to evaluate the ocean front＇s degree of uncertainties and the precision of the representation. It has been proven that the method is not only capable of representing ocean fronts＇ uncertainties, but also provides a new theory and method for uncertainty representation of other oceanic phenomena.
