The Application Of Intelligent Computing In Clustering Analysis

Intelligent algorithms is to simulate the natural biological breeding, foraging, nesting and other acts to solve complex optimization problems method, by virtue of its simple iterative process, and efficient solution efficiency, the excellent performance of the algorithm has been widespread concern of scholars. Clustering is the fragmented data vectors in accordance with a cluster aggregation rules, so that the same categories of data clustered into a group, finally obtained a number of different groups in the process. Clustering in all areas of real life have a universal application. In order to respond to different clustering problem, researchers have developed a variety of possible algorithms. However, emergence of large and complex data sets clustering technology put forward higher requirements, which requires clustering algorithm is scalable, can handle different types of data, it is capable of handling high-dimensional data and the like. Faced with these problems and requirements, the traditional clustering algorithms and other technologies to further improve the clustering performance has become a trend in the current study.This dissertation mainly studies the intelligent algorithm for clustering problems. First, in order to enhance the global search ability of the algorithm to accelerate the convergence speed and improve the accuracy of the algorithm for solving the traditional intelligent algorithm has been improved; secondly, intelligent algorithms and classical improved k-means clustering algorithm fusion; then, the fusion algorithm used in solving the problem of clustering them; Finally, experimental verification and clustering analysis. The experimental results on the classic test case shows that clustering problem solving, intelligent algorithm proposed compared with traditional clustering algorithm has better performance in some cases. The main work is as follows:(1) Improved the PSO, and inertia weight increasing Particle Swarm optimization, fast convergence to ease the optimal value is easy to fall into local optimum conditions, the ability to enhance the global search particle swarm optimization.(2) Improved the gravitational search algorithm, introduced the idea of particle swarm algorithm to learn the optimal solution, search algorithm proposed gravity accelerated learning, speed up the convergence rate, improve the accuracy of the algorithm for solving.(3) The improvements are the two algorithms, bacterial foraging algorithm and k-means algorithm fusion, which is applied to the clustering problem.