欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
仿射传播聚类有与FCM、K-均值聚类方法等效的聚类功能、并且不需要事先确定初始聚类中心．但它也有一个缺陷就是有时在聚类过程中聚类的数目会出现摇摆，最终影响聚类结果．针对此提出了动态调整阻尼系数以消除摇摆及当调整方法失效时自动跳出摇摆等技术，成功改进了仿射传播聚类算法．并把改进后的仿射传播聚类应用于自适应文本模糊聚类中，取得了较好的聚类效果．
英文摘要：
Affinity propagation （AP）clustering has a same ability with FCM, K-means clustering and without selected the initial cluster center and the cluster number, but AP clustering has a limitation：the number of cluster often swings when clustering even impact the result of convergence. For the paper proposes an improved Affinity propagation （imAP） clustering to overcome this limitation suecessfully,that dynamic adjustment damping factors eliminate sway and escaping from sway when the adjustment method failed. Applying imAP to adaptive text fuzzy clustering. The results of experment indicates that imAP obtains good clustering effect, and indicate the results of the experiments.
同期刊论文项目
同项目期刊论文
期刊信息
位置：