Development And Application Of SOFM Hybrid System Based On Fuzzy Computing

The birth of data mining verifies the value and significance of the era of big data,and the cluster analysis technology is particularly important.With the development of global "cloud computing","Internet of Things","artificial intelligence" and other scientific technologies,clustering algorithms have penetrated into all walks of life from residents' lives to scientific research.In traditional industries,a series of clustering studies from land topography surveys to geological research to environmental zoning to waters,plants,air,etc.;in medicine,the use of self-organizing feature mapping models for spectral image separation,cell image recognition and other studies Up to now,the artificial intelligence society is affected by the rapid development of the mobile Internet and its prosperity in all walks of life.Cluster analysis is also increasingly sought after,and in turn has a profound impact on the industry.The essence of clustering is to analyze the samples of the data set and classify similar samples into one class for further processing.Such behavior is widely required in all walks of life.Self-organizing feature map neural network,as a kind of intelligent clustering method,is especially valued because of its self-organizing and adaptive intelligent characteristics.Compared with the traditional SOFM algorithm,the fuzzy SOFM algorithm is based on this new addition to the concept of membership in fuzzy mathematics.The results obtained by traditional clustering algorithms are too regional.At this time,they need to blur their boundaries through fuzzy mathematics.This article has carried out research,discussion and practice through this concept.In the end,the research results of this subject mainly include the following aspects:(1)The basic principles and background of traditional clustering algorithms are studied,and the current status of clustering algorithms is analyzed,and the classification is based on current usage;(2)In-depth research on the basic principles of fuzzy theory,and analyzes some of the more important knowledge points in the current fuzzy theory,such as fuzzy subsets and fuzzy matrices.And gave a brief explanation of the current method of fuzzy cluster analysis;(3)The SOFM algorithm in the clustering algorithm is analyzed in detail,and the membership function in fuzzy mathematics is combined with the traditional SOFM algorithm,a SOFM system based on fuzzy mathematics is proposed,and an internal encryption of an Internet company is used Of users 'real data was verified.The SOFM algorithm based on fuzzy mathematics proposed in this paper has verified its effectiveness in solving the "dead neuron" problem,the ability to adjust the number of clustering categories,and the ability to flexibly select the results in actual use.