基于巢模板的核空间蚁群聚类算法

为了改进蚁群算法因大量引入随机机制所引发的不稳定性,引入巢模板来改进聚类规则,提出一种基于巢模板的核空间蚁群聚类算法,并与原空间上的聚类算法进行比对。该算法用支持向量机的非线性映射函数把数据样本映射到核空间,再用巢模板记忆蚁群群体特征。核空间上的巢模板蚁群聚类算法能较好地处理特征复杂、类别多的数据集,其聚类结果比较接近真实情况,并且效果明显优于原空间上的聚类算法。

Nest Template-Based Ant Clustering Algorithm in Kernel Space

If the features of data samples＇ are complex and with more categories,the ant clustering results are not satisfied.After the analysis of the main reasons,an idea that maps the data samples to kernel space by SVM′ nonlinear mapping function is proposed.The features of data samples are recombined and highlighted in kernel space.The ant clustering algorithm is designed in kernel space and the nest template is been used to improve the stability and accuracy of algorithm.Experimental results on UCI datasets show that the clustering results of nest template ant clustering algorithm in kernel space are closer to the reality.The algorithm can proceed datasets which are complex and with more categories and the result is better than that in original space.