论文部分内容阅读
聚类分组数的自动确定是谱聚类算法中一个亟待解决的问题.针对谱聚类算法聚类分组数的获取问题,提出一种基于人工免疫的自适应谱聚类算法.该算法通过模拟抗体的克隆选择机制和免疫系统的初次免疫应答、二次免疫应答机制,实现了数据样本聚类分组数的自动调整,解决了聚类算法需要人工输入聚类分组数的弊端.并分别在线性模拟数据、非凸模拟数据和UCI数据集上验证了算法的可行性、算法在非凸数据集上的优势以及算法的有效性.实验结果表明该算法可以自动获取正确的聚类分组数,提高聚类效果,减少达到全局最优解时的迭代次数,具有较高的稳定性.
The automatic determination of clustering group number is an urgent problem to be solved in spectral clustering algorithm.Aiming at the problem of obtaining clustering group number of spectral clustering algorithm, an adaptive spectral clustering algorithm based on artificial immune is proposed.This algorithm simulates Antibody cloning selection mechanism and the immune system’s primary immune response, secondary immune response mechanism to achieve the automatic adjustment of the data sample clustering grouping number, to solve the drawbacks of the clustering algorithm need to manually enter the clustering grouping number, and respectively in the linear Simulation data, non-convex analogue data and UCI data set to verify the feasibility of the algorithm and the advantages of the algorithm on non-convex data sets and the effectiveness of the algorithm.The experimental results show that the algorithm can automatically obtain the correct clustering group number, The clustering effect reduces the number of iterations when reaching the global optimal solution and has higher stability.