基于模拟退火改进的自组织映射网络聚类

Improved Self-organizing Mapping Network Clustering Based on Simulated Annealing

自组织映射网络(SOM)具有良好的自组织性和可视化等特征,因此常被用于无监督聚类中.传统的SOM网络在聚类时容易受控制参数和数据集输入顺序的影响,导致聚类的过度拟合和死节点的出现.自组织特征映射网络邻域调整过程中引入模拟退火算法,以一定的概率激活邻域外节点,有效避免网络的过度拟合和死节点的出现.在数据集上的实验证明,引入模拟退火算法SOM网络在一定程度上要优于原始算法.

Self Organizing Mapping network(SOM)has the characteristics of good self-organization and visualization,so it is often used in unsupervised clustering.Traditional SOM networks are easily affected by control parameters and data set input order,resulting in over fitting of clustering and the emergence of dead nodes.Therefore,the simulated annealing algorithm is introduced into the neighborhood adjustment process of the traditional self-organizing feature mapping network to activate the nodes outside the neighborhood with a certain probability,so as to effectively avoid the over fitting of the network and the emergence of dead nodes.Experiments on data sets show that the SOM network with simulated annealing algorithm is better than that of the original algorithm to a certain extent.