Fast clustering method based on multipoint tension model

Abstract

The invention relates to a fast clustering method based on a multipoint tension model, which belongs to the technical field of clustering analysis. In an initial phase, a certain data set is selected from a UCI data set library, and whether the current bees are leading bees is judged; a better solution is found according to an operator, and the algorithm thus jumps out of the local optimum; whether all bees complete a searching activity is judged, and after all bees complete the searching activity, the fitness of all nectar sources currently is evaluated, and the current best solution is recorded; whether the algorithm meets an end condition is judged, and when the algorithm meets the end condition, the optimum solution is outputted, and the algorithm is ended. In a local searching phase, in combination with a crossover operator and a mutation operator in a genetic algorithm, the algorithm convergence speed and the solution diversity are improved, the method is applicable to different scales of data sets with multiple kinds, and the scalability and the robustness are strong.

Description

Technical field [0001] The invention belongs to the technical field of cluster analysis and relates to a rapid clustering method based on a multi-point tension model. Background technique [0002] Cluster analysis is an important unsupervised learning method, which plays an irreplaceable role in identifying the internal structure of data. Clustering is based on the similarity of data objects, dividing the unknown data set into different classes or clusters, so that the data objects in the same cluster have the greatest similarity, and the data objects between different clusters have the least similarity. Cluster analysis has been widely researched and applied in machine learning, pattern recognition, data mining, image processing and other fields. [0003] K-means algorithm is one of the most classic clustering algorithms. It is simple, efficient, and has good local search capabilities, and has been widely used and researched. However, there are disadvantages such as too much dep...

AI Technical Summary

Problems solved by technology

Many scholars have made different improvements to the selection of the initial clustering center, which has improved the clustering effect to a certain extent and improved the quality of the clustering results of the algorithm, but it may still fall into a local optimum.

Images