Gaussian mixture model searching method based on immune clonal selection algorithm

Abstract

The invention provides a Gaussian mixture model searching method based on an immune clonal selection algorithm. Immune clonal selection optimization is used as a basic searching algorithm. Dynamic neighborhood information is used to guide the search of a Gaussian mixture model in a search process, and finally a Gaussian mixture model which can express data characteristics is acquired. The method has the advantages that the searching method is a group-based intelligent random searching method, and has the advantages of group cooperation and heuristic multi-directional search; the method maps the Gaussian mixture model to an antibody, makes full use of the global search advantage of the immune clonal selection algorithm, and uses a mixed clonal mutation policy to improve the search quality; the method provided by the invention integrates the dynamic neighborhood information with anti-noise ability, so as to timely guide the search of the Gaussian mixture model; and the acquired high quality mixture Gaussian model can be used for new data prediction, classification and the like.

Description

technical field [0001] The invention relates to the field of computer information technology, in particular to a method for searching a Gaussian mixture model based on an immune clone selection algorithm. Background technique [0002] Cluster analysis can discover hidden patterns or useful knowledge in data, and it is a very important data mining technique. Simply put, cluster analysis is to divide the target data set into several clusters (groups) according to a certain similarity measure, so that the similarity within the cluster is as large as possible and the similarity between clusters is as small as possible. Cluster analysis technology has been applied in business, biology, geography, insurance industry and Internet and other fields. [0003] Driven by the pursuit of research and the needs of practical applications, cluster analysis techniques have also been developed, and many different clustering techniques have emerged one after another. Gaussian mixed model (GMM...

AI Technical Summary

Problems solved by technology

However, the above search process has the following disadvantages: 1) The EM algorithm belongs to the gradient descent search method, which is easy to fall into a local optimum and has poor global optimization ability; 2) the correlation information between sample points is not considered in the iterative search process: The calculation is considered on a data point basis

These shortcomings make the quality of the obtained GMM lower, hindering the application in practice

Images