Application of Fuzzy Recognition to Model Selection

Abstract

Model selection relies on the attributes of models heavily. And the attributes of models may be certain or uncertain, so how to process these two kinds of attributes, and how to compare the similarity between the object problem and models in term of the attributes are the key issues in model selection. To solve the problem, a new method based on fuzzy recognition is introduced in this article. Firstly, object problem and models are processed by using fuzzy theory. Then, a fuzzy similarity algorithm, which combines advantage of improved index method and that of max-min method, is proposed to select the most appropriate model. Finally, an illustrative example is given to demonstrate validity and rationality of the method.