Incremental F-parallel Attributes Reductions Of Decision Information System

Rough set theory is the method of studying uncertainty,incomplete knowledge,it is widely used in data mining,artificial intelligence,pattern recognition,many other fields.Attribute reduction is a reduction without affecting the quality of classification of original decision table,with minimal attributes to represent the information of decision table.It is one of the core of rough set theoretical research.At present,the parallel attribute reduction is a hot research topic in attribute reduction of rough set,it generalized the rough set theory from a single information table or decision table to multiple decision tables.The idea is more accord with thinking habit of solving problems of human beings,it also fully embodies the idea of granular computing.Because of many advantages,parallel attribute reduction got the attention of many scholars.At present,the incremental attribute reduction algorithm based on decision information system is generally adopted in the following ways:the decision table will be redrawn a new equivalence class when the original decision table is with an new object.In order to reduce the complexity of incremental data processing,an incremental F-parallel attribute reduction algorithm for decision information system is proposed in this paper.The main content of this paper includes four parts:First,the related knowledge of rough set theory,F-rough set,parallel attribute reduction is introduced.Second,based on the F-rough set model and parallel attribute reduction,an incremental parallel attribute reduction algorithm for decision information system is proposed in this paper.One or several new information system was composed for parallel computing with multiple rows which is new in this algorithm.It can adapt well to the new data,avoid the problem of the high time complexity of incremental attribute reduction using heuristic information meanwhile,the time efficiency is improved.Third,the feasibility and efficiency of incremental F-parallel attribute reduction algorithm are illustrated after the time complexity analysis of dynamic attribute reduction algorithm based on uncertain information system and incremental F-parallel attribute reduction algorithm.The five UCI data sets were simulated and compared under the environment of MATLAB R2010a,the result is that the reduction length on some data sets which is obtained from incremental F-parallel attribute reduction algorithm is less than that obtained from dynamic attribute reduction algorithm based on uncertain information system,running time of incremental F-parallel attribute reduction algorithm is also less than dynamic attribute reduction algorithm based on uncertain information system.Thus the efficiency of incremental F-parallel attribute reduction algorithm is proved furtherly.Fourth,the incremental F-parallel attribute reduction algorithm is applied in determination of the types of breast cancer cells,attribute reduction and analysis on Wisconsin breast cancer data set were made and breast cancer patients were diagnosed using incremental F-parallel attribute reduction algorithm,eventually the corresponding decision rules were extracted.Finally,the relationship between the characteristics of breast cancer patients and diagnosis of breast cancer were analysed according to the decision attribute value.