Local Fuzzy Rough Sets And Its Reduction

Attribute reduction is an important application of rough set model,which is widely used in pattern recognition,knowledge discovery and data mining.Similarly,attribute reduction plays an important role in the discussion of fuzzy rough information system.Therefore,the discussion on attribute reduction methods of fuzzy rough information systems has attracted extensive attention.First of all,this paper is discussed in the local fuzzy rough information system.In order to improve the precision of reduction and extend the application range,we extend the local fuzzy rough set model to the cases of double quantization and two universes,and then analyze the attribute reduction methods of these models.The specific research contents are as follows:(1)We propose the local fuzzy rough set and its attribute reduction method.Based on the general fuzzy rough set granular structure,the local fuzzy rough set granular structure is analyzed by proposing the local theory and endowing it with the concept of fuzzy set inclusion degree.First,the local fuzzy rough set is defined,then the corresponding properties of the model are discussed,and the corresponding attribute reduction algorithm is proposed for the local fuzzy rough set model proposed previously.Finally,its efficiency and accuracy are verified by corresponding examples.(2)The idea of local fuzzy rough sets is extended from the perspective of granular structure to information entropy.We analyze the rationality of the definition of local fuzzy information entropy in fuzzy rough sets,and introduce the definitions of local fuzzy information entropy and other corresponding information entropy.Then propose attribute reduction methods based on local information entropy according to the definition,and further analyze the accelerated forms of these reduction algorithms.Finally,the rationality and effectiveness of the method are verified through examples and experiments.(3)In order to solve the problems of the lack of full supervised data and the low computational efficiency that may exist in the grade fuzzy rough set model,the concept of local be used in the grade fuzzy rough set model to generate the local grade fuzzy rough set model.Secondly,considering the influence of double quantization information,in order to improve the accuracy of single quantization model,the local double quantization fuzzy rough set model is generated by combining the corresponding local fuzzy rough set model and local grade fuzzy rough set model.Then the corresponding attribute reduction method is proposed,and its rationality and effectiveness are verified by an example.(4)In order to expand the application scope of the local fuzzy rough set model,the local fuzzy rough set model over two universes is proposed,and its properties and decision rules are further discussed.In addition,considering the application of practical problems,a reduction method of the corresponding model is proposed.Finally,through experimental verification,the proposed method has good classification performance and decision-making results.Finally,some problems related to local fuzzy rough set model are listed for further study.