摘要
为了提高混合聚类算法的准确率,提出基于优化人工鱼群算法的混合聚类算法。引入人工鱼群算法,辅以鲁棒性更强的K中心点算法优化了混合聚类方法的聚集效果。通过对人工鱼的行为和参数进行改善,避免了聚类效果易受离群点影响的问题,对理噪声数据的处理更好。结合K-中心点算法与人工鱼群算法的优势,解决了聚类算法初值依赖性,克服了鱼群算法后期迭代速度慢问题。仿真结果表明,该算法全局优化性能稳定,收敛速度加快,聚类效果明显提高,获得了较优的中心点与清晰地聚类划分。
To improve the accuracy of hybrid clustering algorithm, a novet ctustermg aigorithm based on optimized artificial fish swarm algorithm is proposed. By introducing artificial fish swarm algorithm and employing K-medoids algorithm which has stronger robustness, aggregation results of the method are optimized. Through the improvement of artificial fish behavior and the parameters, it is designed to avoid susceptibility of the algorithm affected by the outliers and make the effect of noise data processing better. Combined the advantages of K-medoids and artificial fish swarm algorithm, the algorithm not only solves the initial dependence, but also overcomes the slower iteration rate in later period. Finally, the simulations prove that performance of proposed approach is stable, whose convergence speed is fast and the clear divisions are obtained.
出处
《计算机工程与设计》
CSCD
北大核心
2014年第3期1041-1045,共5页
Computer Engineering and Design
基金
国家科学技术部批准的国际科技合作与交流专项基金项目(2011DFA91810-5)
国家教育部批准的新世纪优秀人才支持计划基金项目(NCET-12-1012)
关键词
混合聚类
人工鱼群算法
优化参数
K-中心点算法
聚类效果
hybrid clustering
artificial fish swarm algorithm
optimized parameters
K-medoids algorithm
clustering effect