Research On Attribute Reduction Algorithms Based On Extended Rough Set Model

Rough set theory is an important mathematical tool for dealing with imprecision,incompleteness and uncertainty.The classification condition of the classic rough set is a strict indistinguishable relationship,but it can not directly process mixed data and accurately express some fuzzy concepts in life,resulting in certain limitations in practical application.Therefore,the extended model combining rough set with other mathematical theories have been appeared,and it is widely applied to the solution of practical problems.Attribute reduction,as one of the core contents of theoretical research on extended model of rough set,plays a key role in dealing with the uncertainty problem in information time.It can eliminate redundant or irrelevant attributes from the original attributes set and improve the classification efficiency of the model while maintaining the classification ability.In this paper,our work focus on the defects of the strict equivalence relation of classical rough set model will lead to information loss when discretizing mixed data in practical application,and the inability to express the fuzzy degree of decision-making problems,the attribute reduction algorithms based on the extended rough set model are studied.The main research work of this paper is as follows:1.Since the inability of classical rough set model to deal with multi-label mixed data directly and the importance of each label is different in label space,we propose a novel multi-label attribute reduction algorithm based on neighborhood rough set.This model uses the difference between the average distance from a given sample to its heterogeneous sample and the average distance from the sample to its homogeneous sample in attribute space to give weight to labels,mining the importance of labels.Then,we propose a new method of positive region calculation under the framework of neighborhood rough set,which effectively improves the calculation of positive region.Moreover,the proposed method overcomes the shortcomings of classical rough set model needs discretization operation to process mixeddata and lose important information.Extensive experiments show that the classification performance of proposed method is better than other multi-label attribute reduction algorithms.2.Considering the high dimensionality of multi-label data and the complexity of data types interfere with classification performance.We develop an algorithm called a multi-label attribute reduction based on variable precision fuzzy neighborhood rough set.Our proposed method first uses the parameterized fuzzy neighborhood granule to define multi-label fuzzy decision and decision classes,and then defines a multi-label variable precision fuzzy neighborhood rough approximation space based on inclusion of rough set,which solves the rigor of upper and lower approximate calculation of classical rough set model.Extensive experiments on different types of data sets demonstrate that the proposed method can effectively reduce the dimension of attribute space and achieve higher prediction performance compared with several popular multi-label attribute reduction algorithms.3.Classical rough set can not express the degree of acceptance,objection and hesitation of decision-making problems in life simultaneously,as well as the problem that the number of data attributes dimension and the increase in decision-making time in practical applications in the era of big data.We propose a single valued neutrosophic rough set attribute reduction algorithm based on inclusion theory to reduce the dimension of attribute space.Firstly,the inclusion degree formulas and related concepts of the single valued neutrosophic rough set are defined;Secondly,the variable precision parameters is added to make the practical decision-making problems more reasonable.Finally,a specific example is gave to illustrate that the proposed algorithm can reduce the time consumption of decision-making process and deal with uncertainty information.In addition,it makes the single valued neutrosophic rough set more practical for solving decision problems.