遗传模拟退火算法在k—means聚类中的应用

Research of K-means Clustering Method Based on Genetic Simulated Annealing Algorithm

针对传统的k-means算法执行的结果过度依赖与初始聚类中心的选择，容易陷入局部最优解，全局搜索能力不强等缺点，该文提出将遗传算法（GA）与模拟退火算法（SA）相结合的混合遗传算法，此方法比传统的遗传算法具有更好的全局收敛性，避免了早熟的问题。将此算法应用到k-means聚类领域中，可以有效的克服k-means算法的缺陷，同时优化聚类的结果。实验表明该方法是一种高效精确的聚类方法。

As K-means Clustering Algorithm is excessively depend on the choice of the initial cluster centers, it leads to be involved in lo- cally optimal solution and does not have comprehensive ability to search. We propose a method based on Genetic algorithms and Simulated annealing called Genetic Simulated annealing algorithm. The new method has better Global convergence compared with Genetic algo- rithms. It also overcomes the aspect of premature convergence problem. So we can apply this algorithms to the K-means Clustering Meth- od so as to overcome defect of K-means Clustering Method and optimize the result of clustering . Experiments indicated that this algo- rithm is efficient and accurate.