Font Size: a A A

Fusion Ant Heap Of Clustering And Fuzzy C-Means Clustering Algorithm Research And Analysis

Posted on:2013-03-01Degree:MasterType:Thesis
Country:ChinaCandidate:F ZhouFull Text:PDF
GTID:2248330371997863Subject:Computer software and theory
Abstract/Summary:PDF Full Text Request
Data mining is a technology that mines useful information from complex data set of human society since the twentieth century into the era of information explosion, a large number of data generated surge in the amount of information brought about rapid development of commercial, cultural and science and technology and which brings problems to obtain useful information foe humans at the same time, so data mining in this form After years of efforts of the scholars, data mining technology becomes more mature, more extensive range of applications and multiple branches, the cluster analysis is one of the best.Cluster analysis is more successful in data mining research and application of a mining technology. Applications in the real world, diversity and complexity of data to the cluster analysis a great deal of trouble, a certain type of data often correspond to a specific algorithm, most algorithms don’t have universal application.In recent years, according to reasearch ant colony behavior scholars raised clustering algorithm, as opposed to traditional cluster analysis algorithm that the algorithm can handle the data type of diversity, and can find clusters of arbitrary shape without too more human guidance, the robustness of the algorithm is higher.But as a simulation of evolution.algorithm is very obvious defects and runs takes a lot of time.Therefore, this article attempts to combine it with a mature cluster analysis algorithm in order to achieve better clustering results through two complementary advantages In this article, I select the traditional clustering algorithm—the fuzzy C-means clustering algorithm.Fuzzy clustering algorithm is a very mature cluster analysis algorithm, that is based on function optimization method and made use of calculus in the mathematical theory of computing technology.It takes a more objective manner about reflection of the real world, then makes the fuzzy clustering is widely used in image segmentation, pattern recognition, large-scale data analysis There are many types about fuzzy clustering algorithm, and the fuzzy C-means FCM algorithm, these algorithms are more representative and also has a very wide range of applications and very successful in society.The algorithm is simple, fast running speed, but in the initial of algorithm have to set the number of clusters and cluster centers parameters which led to the algorithm by man-made a great impact.The hybrid clustering by studying the different clustering algorithms and two or more clustering algorithm combination of the algorithm, so that can get more excellent clustering effect.This article ant heap clustering algorithm and fuzzy C-means clustering algorithm to combine a series of improvements, and the original ant heap algorithm, in order to adapt to the new hybrid clustering algorithm.The improvements of ant heap algorithm mainly includes three parts are for the defects of the algorithm is time-consuming.First, the algorithm for deletion of the three speed settings of the original algorithm and retaining a kind of speed,taht in order to be completed as soon as possible to the initial cluster. Second, the new algorithm keeps the location of the data objects in the two-dimensional array so that the ants find the data object can be avoided empty jump and save time.Finally, the strategy of ants unload data object has been modified to fit to the data on the appropriate location.New algorithm for the last publicly available data for the UCI Machine Learning Repository experiment, the experimental results show that this new hybrid algorithm is effective to improve the quality of clustering results. However, the algorithm still takes a long time, mainly present in the ant heap clustering part,although this article has a series of improvements to the ant heap clustering relative to the time efficiency of the original ant heap clustering algorithm has been improved, but the algorithm still has room for improvement.The next step is for the ant heap clustering algorithm convergence speed of shortcomings for further improvement, so that the algorithm has better efficiency, and attempt to put the other idea of the algorithm into which explore the new way.
Keywords/Search Tags:Ant Colony Algorithm, Ant heap clustering, Artificial ants, Fuzzy C meansclustering, Hybrid Clustering
PDF Full Text Request
Related items