Fuzzy Classification Algorithm Based on Mutually Restraint

摘要

On the basis of most steady state structure of rigid bodies in physical theory, this paper regards data set as physical object, and proposes a dummy digital substantiation viewpoint and some conceptions such as representative data five element groups and model restraint. Then, the algorithm of elasticity based four-point center and borderline (EFCBA) and the algorithm of based on mutual restrict fuzzy data classification (MRFCA) are presented. They have the main features as follows: l)To combine data set with some concepts of physics, they put forward to the thinking of data substantiation, and lay the foundation of getting representative data of irregular figures and unknown boundary sets; 2) They alleviate the effect caused by irregularity of figures to getting set determination features, and enhance the representation of representative data in set; 3) Classification algorithm has a strong power in classifying, and it has a strong ability to identifying unknown data as well as a strong selfadaptive. The experiment shows the EFCBA enhanced data represented, alleviated the graphics irregularity's influence; The MRFCA used the characteristic of represent data effectively, enhanced the classification's accuracy, and had the good adaptive ability and unknown data classification ability.