Fuzzy And Neural Network-based Clustering Analysis

Along with the science of society developing, the information technology not only developed but also more and more perfection. Internet and Web data processing technology is also be popular, and the access method diversified. With the information data increasing continuously, the internal structure of the data draw people more attention to find the solution.The cluster analysis which is an important research technique was used to classify the object with mathematics methods. As it is said "The human by the classified collection, the thing by the group minute". The cluster analysis is an important methods in the knowledge discovery and machine learning as well as in the middle of data mining. The cluster analysis is divide identical kind of data characteristic similar as far as possible which is do classification only according to the similarity of data, and without the apriority knowledge premise, therefore the cluster analysis belongs to one non-surveillance's pattern recognition method in the pattern recognition area.So it is needed to establish to pattern recognition's research may distinguish automatically the data-in category information the taxonomy. The non-surveillance's cluster analysis algorithm may unearth the data latent organization of information, by widespread application each data analysis domain, including computer vision analysis, imagery processing, biological science, social sciences as well as medical information and so on.This paper main detailed cluster analysis of fuzzy relationship and neural network, as well as based on Takagi-Sugeno fuzzy system's fuzzy neural network pattern classification. And aims at the three dimensional input pattern sample, selects some pattern taxonomic approach, and classify the pattern sample through mathematics software MATLAB programming. The paper prime task is as follows:1. Introduced cluster analysis's algorithm and the application, summarized the theory level of cluster analysis and the theory background, and have elaborated the fuzzy data relationship and neural network's cluster analysis's algorithm.2. Selected some pattern taxonomic approach, with the help of MATLAB software, realized the three dimensional input space pattern classification. 3. Because of the flexibility of ANFIS system, also the MATLAB software formidable computation and the simulation function,we have constructed a neural network using ANFIS which based on Takagi-Sugeno fuzzy system's, has realized the three dimensional pattern classification.