基于流形距离的量子进化聚类算法

Quantum-Inspired Evolutionary Clustering Algorithm Based on Manifold Distance

基于量子计算的机理和特性,并结合进化计算,本文提出了一种新颖的量子进化聚类算法(QEAM),在该聚类算法中引入了一种新的距离测度函数——流形距离.新方法将聚类归属为优化问题,通过运用量子进化的机理更快地搜索到最优聚类中心,从而得到最优隶属度矩阵划分;同时,通过基于流形距离的相似性度量,有效利用样本所具有的全局一致性信息,充分挖掘样本的空间分布信息,对样本进行正确的类别划分.将本文算法(QEAM)与基于流形距离的免疫进化算法(IEAM),遗传聚类算法(GAC)以及模糊C-均值算法(FCM)进行了性能比较,对6个人工数据集和3个UCI数据集的仿真实验结果显示,QEAM对样本空间分布复杂的聚类问题具有较高的准确率和较好的鲁棒性.

Based on the concepts and principles of quantum computing,a novel quantum-inspired evolutionary algorithm for data clustering（QEAM） is proposed in this paper by using a novel distance measurement index called manifold distance which can measure the geodesic distance along with the manifold.The clustering problem is viewed as an optimization problem.Our main motives of using QEAM consist in searching for appropriate cluster center by using the principles of quantum evolutionary computation,so that a similarity metric of clusters are optimized more quickly and effectively.The experimental results on six artificial datasets and three UCI datasets show the superiority of QEAM over an immune evolutionary clustering algorithm with manifold distance（IEAM）,a genetic algorithm for clustering（GAC） and fuzzy c-means algorithm（FCM）.