The Research On The Complicated Classfication Problem And Fuzzy Set Reclassfication Model

The problems of classification often occur in our daily life, social activity, scientific research and production, study and work. In many fields, the problems of classification are all main problems which need to be solved, such as clinical diagnosis in medicine, judgement of the machine condition in industrial production, intelligent voice recognition, and so on. And the problem of classification has become the most principal and the most difficult problem to solve. This article puts forward a model (Fuzzy Set Reclassfication Model, FSR Model), which takes advantage of the combination of fuzzy theory, machine learning and artificial intelligence algorithm to solve complicated problems. And the point of FSR is taking advantage of the knowledge of fuzzy mathematics to simplify the complicated problems to the easy ones.The theory of fuzzy takes advantage of the conception of subjection function to measure the adscriptional degree of data. The adscription of data can be judged more correctly by reasonable subjection function. Usually, the defination of subjection function needs enough knowledge of apriority, and it has large inaccuracies.The Genetic Algorithms is a developing arithmetic, which borrows natural selection and evolution in biology. It has the features of high parallel, random and strong adaptability, so it can solve the problem of large search space effectively.It can solve the problems of classification by taking advantage of fuzzy theory. But in the premise of not getting enough knowledge, the scale of parameter to be assured is very huge. And the Genetic Algorithms can search the space effectively. Taking advantage of the Genetic Algorithms to optimize subjection function can get better effect and correctness. The FSR Model is based on it, and it reduces the demand of accuracy of subjection function further. FSR Model regulates that samples are correct classified which only need to be distributed into one of the correct group or the closer group.The focuses of this article are putting forward the FSR model and explanation of the structure of the FSR model. And it ensures the viability of the entire model, by using many groups of data set of different type to carry out performance testing for every module.