论文部分内容阅读
一个实际的聚类问题中,各维属性的贡献通常是不一样的,具有主次之分,但传统的聚类算法将所有属性赋予相同的权重.如果能够将其重要属性赋予较大属性权重,则可以提高聚类效果.采用改进粒子群优化算法为每一维属性求取相应权重,并将得到的权重应用到迭代自组织数据分析技术算法中,构建一种基于改进粒子群属性权重的迭代自组织数据分析技术算法.试验结果表明,合理的权重改善了聚类算法的性能,提高了聚类质量.
In an actual clustering problem, the contribution of each dimension attribute is usually different, with primary and secondary points, but the traditional clustering algorithm assigns the same weight to all the attributes.If we can assign its important attributes to the larger attribute weights , Then the clustering effect can be improved.An improved weight swarm optimization algorithm is used to obtain the corresponding weight for each dimension attribute and the weight is applied to the algorithm of iterative self-organizing data analysis to build a new algorithm based on the improved weight of the PSO Iterative self-organizing data analysis algorithm.The experimental results show that the reasonable weights improve the performance of clustering algorithm and improve the clustering quality.