RDPSO算法与K-Means聚类算法相结合的混合集群技术
|
摘要
K-Means算法初始值选取不当容易局部收敛,引入一种随机漂移粒子群优化算法(RDPSO)与K-Means算法相结合的混合集群技术。RDPSO算法是一种具有较强全局搜索能力的粒子群算法,增强了粒子群优化算法的性能,此算法优秀的全局搜索能力优化了初始聚类中心的选取,K-Means算法易于搜索到全局最优的初始值,从而获得全局最优解。与PSO、K-Means的混合算法相比,通过实验分析表明了文中提出的混合集群技术能够更有效地加快收敛速度,提高了全局搜索能力。
Aiming at the shortcoming of the traditional k-means algorithm's improper selection of the initial clustering center and easy local convergence,this paper proposes a hybrid clustering technology combining the random drift particle swarm optimization algorithm( RDPSO) and the k-means clustering algorithm. RDPSO algorithm is combined with free electrons in the metal conductor orientation drift motion and stochastic random thermal motion mode is put forward a kind of strong global search ability of particle swarm algorithm. To enhance the performance of the PSO algorithm,a strong global search ability has optimized the selection of the initial clustering center for K-Means algorithm to search the global optimal initial clustering centers which is easy to obtain the global optimal solution. Compared with PSO and k-Means hybrid algorithms,the experimental results show that the hybrid clustering technology proposed in this paper can accelerate the convergence rate more effectively and improve the global search ability.
Aiming at the shortcoming of the traditional k-means algorithm's improper selection of the initial clustering center and easy local convergence,this paper proposes a hybrid clustering technology combining the random drift particle swarm optimization algorithm( RDPSO) and the k-means clustering algorithm. RDPSO algorithm is combined with free electrons in the metal conductor orientation drift motion and stochastic random thermal motion mode is put forward a kind of strong global search ability of particle swarm algorithm. To enhance the performance of the PSO algorithm,a strong global search ability has optimized the selection of the initial clustering center for K-Means algorithm to search the global optimal initial clustering centers which is easy to obtain the global optimal solution. Compared with PSO and k-Means hybrid algorithms,the experimental results show that the hybrid clustering technology proposed in this paper can accelerate the convergence rate more effectively and improve the global search ability.
引文
[1]何清,李宁,罗文娟,等.大数据下的机器学习算法综述[J].模式识别与人工智能,2014,27(04):327-336.
[2]张沛明.魏楠.基于聚类算法的巨量轨迹数据分析[J].鞍山师范学院学报,2017,19(2):54-58.
[3]王勇,唐靖,饶勤菲,等.高效率的K-means最佳聚类数确定算法[J].计算机应用,2014,34(05):1331-1335.
[4]张慧斌,王鸿斌,胡志军. PSO算法全局收敛性分析[J].计算机工程与应用,2011,47(34):61-63.
[5]方伟,周建宏.基于频繁覆盖策略的随机漂移粒子群优化算法[J].控制与决策,2017,32(12):2127-2136.
[6]陶新民,徐晶,杨立标,等.一种改进的粒子群和K均值混合聚类算法[J].电子与信息学报,2010,32(01):92-97.
[7]谢秀华,李陶深.一种基于改进PSO的K-means优化聚类算法[J].计算机技术与发展,2014,24(02):34-38.
[8]魏新红,张凯.一种改进的PSO-Means聚类优化算法[J].河南科技大学学报:自然科学版,2011,32(02):41-43+109.
[2]张沛明.魏楠.基于聚类算法的巨量轨迹数据分析[J].鞍山师范学院学报,2017,19(2):54-58.
[3]王勇,唐靖,饶勤菲,等.高效率的K-means最佳聚类数确定算法[J].计算机应用,2014,34(05):1331-1335.
[4]张慧斌,王鸿斌,胡志军. PSO算法全局收敛性分析[J].计算机工程与应用,2011,47(34):61-63.
[5]方伟,周建宏.基于频繁覆盖策略的随机漂移粒子群优化算法[J].控制与决策,2017,32(12):2127-2136.
[6]陶新民,徐晶,杨立标,等.一种改进的粒子群和K均值混合聚类算法[J].电子与信息学报,2010,32(01):92-97.
[7]谢秀华,李陶深.一种基于改进PSO的K-means优化聚类算法[J].计算机技术与发展,2014,24(02):34-38.
[8]魏新红,张凯.一种改进的PSO-Means聚类优化算法[J].河南科技大学学报:自然科学版,2011,32(02):41-43+109.