Research On Fuzzy C-means

Traditional cluster analysis dataset accurately divided into a class, the so-called hard division. But in fact most things in the property, there is ambiguity, no clear boundaries between things, not either-or nature of the concept of fuzzy clustering is more suitable to the nature of things more objectively reflect the reality. Currently, fuzzy C-means (fuzzy C-means, the FCM) clustering algorithm is the most widely used fuzzy clustering algorithm. The direction of data mining in recent years, many algorithms for the explosive growth of social data, these algorithms varies, and a variety of information on the network is growing every day. In these algorithms, fuzzy c-means clustering algorithm can be considered a good algorithm, have long been proposed, and evolved into many related algorithms.Fuzzy c-means clustering algorithm is the product of cluster analysis and fuzzy theory. Fuzzy theory (Fuzzy Theory) on top of fuzzy sets, is to depict and analyze the human-specific language is ambiguous. And then to introduce the cluster analysis, clustering of things according to certain rules and principles of the classification process. No experience in the classification process, there is no teacher guidance, relies entirely on the similarity between the things to be divided, so it belongs to unsupervised classification areas. The cluster analysis method is to use mathematics to things classification clustering analysis of the four main aspects, including the equivalence relation-based clustering methods, graph theory, clustering method and the objective function-based clustering method, hierarchical clustering method. In addition to the other three methods based on the outside of the objective function clustering method can not adapt to the large amounts of data, which highlights the significance of research-based objective function.As we all know, the improved algorithm of the c-means clustering algorithm, and c-means clustering algorithm is to assume that their influence on the results of the properties of the clustering process is the same. this assumption is that the algorithm is better meaning. But in today’s society, the amount of data is an increasingly large, the complexity of the data is as straight up. The original properties of (?) clustering process assumption has been unable to meet the needs of the present.(?) e face of today’s high-dimensional data, many of the properties is in fact a noise data, the results will play a counter-data. This property also may be no effect on the results of these attributes, we can not simply assume it into the same attributes as the results. Because of the clustering results of different attributes in the data have different roles, each attribute may have different purposes. Therefore, understanding and analysis of the data attributes and attributes in the clustering process is particularly important in the present cluster analysis.In this paper, the fuzzy c-means clustering on the basis of weight, semi-supervised point density and study in computer forensics.(1) by adding the weights. The property is worth the weight value is very important in the practical application, this chapter presented a new fuzzy c-means clustering algorithm (New Fuzzy c-means NFCM). The algorithm can make the right of each attribute value, and can not the dominant class structure is extracted.(2) by adding the semi-supervised point density. A group like or relatively large difference in the number of categories of data, fuzzy c-means clustering algorithm at this time does not handle well, that semi-supervised learning of fuzzy c-means clustering algorithm is not a good deal with this problem. They can not be a good division of the data, because they are data such as division, these are the semi-supervised learning of fuzzy c-means clustering algorithm in some aspects of the defect, which prompted to join the semi-supervised fuzzy c-means clustering algorithm weighted production. Semi-supervised point density of fuzzy c-means clustering algorithm in the previous two algorithms based on semi-supervised point-density fuzzy c-means clustering algorithm is based on the data point density of the sample set plus weights calculated, so able drawbacks of the fuzzy c-means clustering algorithm, fuzzy c-means clustering algorithm and the semi-supervised learning data samples were divided into improved, so that we can differentiated divide the data.(3) In the traditional forensic analysis is obtained from the visible data of the known information. Computer forensic analysis is to obtain useful information from vast amounts of various types of electronic data. This process manually is not good, need to rely on computer systems, which filter out evidence of data related to computer crime. This paper mainly applications based on principal component analysis, fuzzy clustering algorithm to cluster in order to more in-depth analysis of electronic evidence. Computer Forensics is the application of computer technology to access, investigate and analyze the technology of computer crime.