正在加载图片...
第6期 朱书伟,等:融合并行混沌莹火虫算法的K-调和均值聚类 ·879- 表5p=3.5时各算法实验结果 [2]ZHANG Bin,HSU M,DAYAL U.K-harmonic means-a da- Table 5 The result of four algorithms when p =3.5 ta clustering algorithm.Technical Report HPL-1999-124 数据集 算法 KHM(X,C)F-measure 时间/s [R].Hewlett-Packard Laboratories,1999. [3]YANG Fengqin,SUN Tieli,ZHANG Changhai.An efficient KHM 109.84 0.892 0.178 hybrid data clustering method based on K-harmonic means KHM-FA 109.53 0.892 1.304 Iris and particle swarm optimization[J].Expert Systems with KHM-PSO 109.61 0.892 1.235 Applications,2009,36(6):9847-9852. 本文算法 109.22 0.892 2.311 [4]ALGUW AIZANI A,HANSEN P,MLADENOVIC N,et al. KHM 2567.9 0.700 0.684 Variable neighborhood search for harmonic means clustering KHM-FA 2558.9 0.700 5.096 Sphere [J].Applied Mathematical Modelling,2011,35 (6): KHM-PSO 2564.4 0.700 5.088 2688-2694. 本文算法 2553.5 0.700 9.662 [5]HUNG C H,CHIOU H M,YANG Weining.Candidate KHM 2.7175×100 0.630 0.273 groups search for K-harmonic means data clustering[J]. KHM-FA 1.4204×100 0.655 2.201 Applied Mathematical Modelling,2013,37(24):10123- Wine KHM-PSO 1.4419×100 0.636 2.140 10128 本文算法 1.4193×100 0.661 3.945 [6]汪中,刘贵全,陈恩红.基于模糊K-harmonic means的谱 KHM 7.0899×10 聚类算法[J].智能系统学报,2009,4(2):95-99. 0.535 0.928 WANG Zhong,LIU Guiquan,CHEN Enhong.A spectral KHM-FA 2.3242×10 0.537 6.667 Image KHM-PSO clustering algorithm based on fuzzy K-harmonic means[J]. 1.6685×10 0.538 6.843 本文算法 CAAI Transactions on Intelligent Systems,2009,4(2): 1.4946×10 0.540 12.162 95-99 KHM 380733.23 0.455 1.966 [7]WU Xiaohong,WU Bin,SUN Jun,et al.A hybrid fuzzy K- KHM-FA 380013.07 0.458 14.892 CMC harmonic means clustering algorithm[J].Applied Mathe- KHM-PSO 380381.58 0.458 14.855 matical Modelling,2015,39(12):3398-3409. 本文算法 379782.74 0.460 26.632 [8]王建峰,孙超,姜守达.基于粒子群优化的组合测试数 KHM 4.1714×10 0.715 12.437 据生成算法[J].哈尔滨工程大学学报,2013,34(4): KHM-FA 4.1630×109 0.721 92.174 477-482 Satellite KHM-PS04.1506×109 0.709 97.873 WANG Jianfeng,SUN Chao,JIANG Shouda.Improved al- 本文算法4.1597×10 0.724 175.75 gorithm for combinatorial test data generation based on parti- cle swarm optimization[J].Journal of Harbin Engineering 4结束语 University,2013,34(4):477-482. 由于传统的KHM算法具有易陷于局部最优解 [9]HE Yaoyao,YANG Shanlin,XU Qifa.Short-term cascaded hydroelectric system scheduling based on chaotic particle 的问题,本文基于一种高效的群智能优化算法提出 swarm optimization using improved logistic map[J].Com- 了一种混合的聚类算法,在KHM中融合了混沌优 munications in Nonlinear Science and Numerical Simula- 化改进的莹火虫算法,不断优化其聚类中心。实验 tiom,2013,18(7):1746-1756. 结果表明,本文算法的综合性能优于KHM以及2 [10]HE Yaoyao,XU Qifa,YANG Shanlin,et al.A novel cha- 种混合聚类算法KHM-FA和KHM-PSO,具有更高 otic differential evolution algorithm for short-term cascaded 的聚类准确性和稳定性,能够有效地避免陷入局部 hydroelectric system scheduling[J].International Journal 最优。但是本文算法的运行时间相对比较长,在数 of Electrical Power Energy Systems,2014,61:455- 据量较大的情况下具有较大的计算开销而影响了算 462. 法的效率,接下来可以针对算法效率的改善开展进 [11]廖煜雷,刘鹏,王建,等.基于改进人工鱼群算法的无 人艇控制参数优化[J].哈尔滨工程大学学报,2014, 一步研究工作。此外,可以尝试将PCLSFA应用于 35(7):800-806 其他的优化问题中。 LIAO Yulei,LIU Peng,WANG Jian,et al.Control pa- 参考文献: rameter optimization for the unmanned surface vehicle with the improved artificial fish swarm algorithm[J].Journal of [1]JAIN A K.Data clustering:50 years beyond K-means[J]. Harbin Engineering University,2014,35(7):800-806. Pattern Recognition Letters,2010,31(8):651-666 [12YANG Xinshe.Firefly algorithm,stochastic test functions表 5 p = 3.5 时各算法实验结果 Table 5 The result of four algorithms when p = 3.5 数据集 算法 KHM( X,C ) F⁃measure 时间/ s Iris KHM KHM⁃FA KHM⁃PSO 本文算法 109.84 109.53 109.61 109.22 0.892 0.892 0.892 0.892 0.178 1.304 1.235 2.311 Sphere KHM KHM⁃FA KHM⁃PSO 本文算法 2 567.9 2 558.9 2 564.4 2 553.5 0.700 0.700 0.700 0.700 0.684 5.096 5.088 9.662 Wine KHM KHM⁃FA KHM⁃PSO 本文算法 2.717 5×10 10 1.420 4×10 10 1.441 9×10 10 1.419 3×10 10 0.630 0.655 0.636 0.661 0.273 2.201 2.140 3.945 Image KHM KHM⁃FA KHM⁃PSO 本文算法 7.089 9×10 9 2.324 2×10 9 1.668 5×10 9 1.494 6×10 9 0.535 0.537 0.538 0.540 0.928 6.667 6.843 12.162 CMC KHM KHM⁃FA KHM⁃PSO 本文算法 380 733.23 380 013.07 380 381.58 379 782.74 0.455 0.458 0.458 0.460 1.966 14.892 14.855 26.632 Satellite KHM KHM⁃FA KHM⁃PSO 本文算法 4.171 4×10 9 4.163 0×10 9 4.150 6×10 9 4.159 7×10 9 0.715 0.721 0.709 0.724 12.437 92.174 97.873 175.75 4 结束语 由于传统的 KHM 算法具有易陷于局部最优解 的问题,本文基于一种高效的群智能优化算法提出 了一种混合的聚类算法,在 KHM 中融合了混沌优 化改进的萤火虫算法,不断优化其聚类中心。 实验 结果表明,本文算法的综合性能优于 KHM 以及 2 种混合聚类算法 KHM⁃FA 和 KHM⁃PSO,具有更高 的聚类准确性和稳定性,能够有效地避免陷入局部 最优。 但是本文算法的运行时间相对比较长,在数 据量较大的情况下具有较大的计算开销而影响了算 法的效率,接下来可以针对算法效率的改善开展进 一步研究工作。 此外,可以尝试将 PCLSFA 应用于 其他的优化问题中。 参考文献: [1]JAIN A K. Data clustering: 50 years beyond K⁃means[ J]. Pattern Recognition Letters, 2010, 31(8): 651⁃666. [2]ZHANG Bin, HSU M, DAYAL U. K⁃harmonic means⁃a da⁃ ta clustering algorithm. Technical Report HPL⁃1999⁃124 [R]. Hewlett⁃Packard Laboratories, 1999. [3]YANG Fengqin, SUN Tieli, ZHANG Changhai. An efficient hybrid data clustering method based on K⁃harmonic means and particle swarm optimization [ J]. Expert Systems with Applications, 2009, 36(6): 9847⁃9852. [4]ALGUWAIZANI A, HANSEN P, MLADENOVIC N, et al. Variable neighborhood search for harmonic means clustering [J ]. Applied Mathematical Modelling, 2011, 35 ( 6 ): 2688⁃2694. [5] HUNG C H, CHIOU H M, YANG Weining. Candidate groups search for K⁃harmonic means data clustering [ J]. Applied Mathematical Modelling, 2013, 37( 24): 10123⁃ 10128. [6]汪中, 刘贵全, 陈恩红. 基于模糊 K⁃harmonic means 的谱 聚类算法[J]. 智能系统学报, 2009, 4(2): 95⁃99. WANG Zhong, LIU Guiquan, CHEN Enhong. A spectral clustering algorithm based on fuzzy K⁃harmonic means[ J]. CAAI Transactions on Intelligent Systems, 2009, 4 ( 2): 95⁃99. [7]WU Xiaohong, WU Bin, SUN Jun, et al. A hybrid fuzzy K⁃ harmonic means clustering algorithm [ J]. Applied Mathe⁃ matical Modelling, 2015, 39(12): 3398⁃3409. [8]王建峰, 孙超, 姜守达. 基于粒子群优化的组合测试数 据生成算法[ J]. 哈尔滨工程大学学报, 2013, 34( 4): 477⁃482. WANG Jianfeng, SUN Chao, JIANG Shouda. Improved al⁃ gorithm for combinatorial test data generation based on parti⁃ cle swarm optimization [ J]. Journal of Harbin Engineering University, 2013, 34(4): 477⁃482. [9]HE Yaoyao, YANG Shanlin, XU Qifa. Short⁃term cascaded hydroelectric system scheduling based on chaotic particle swarm optimization using improved logistic map [ J]. Com⁃ munications in Nonlinear Science and Numerical Simula⁃ tion, 2013, 18(7): 1746⁃1756. [10]HE Yaoyao, XU Qifa, YANG Shanlin, et al. A novel cha⁃ otic differential evolution algorithm for short⁃term cascaded hydroelectric system scheduling [ J]. International Journal of Electrical Power & Energy Systems, 2014, 61: 455⁃ 462. [11]廖煜雷, 刘鹏, 王建, 等. 基于改进人工鱼群算法的无 人艇控制参数优化[ J]. 哈尔滨工程大学学报, 2014, 35(7): 800⁃806. LIAO Yulei, LIU Peng, WANG Jian, et al. Control pa⁃ rameter optimization for the unmanned surface vehicle with the improved artificial fish swarm algorithm[J]. Journal of Harbin Engineering University, 2014, 35(7): 800⁃806. [12]YANG Xinshe. Firefly algorithm, stochastic test functions 第 6 期 朱书伟,等:融合并行混沌萤火虫算法的 K⁃调和均值聚类 ·879·
<<向上翻页向下翻页>>
©2008-现在 cucdc.com 高等教育资讯网 版权所有