Hierarchical clustering driven by cognitive features  

作　　者：LI ChunZhong XU ZongBen QIAO Chen LUO Tao 

机构地区：[1]Institute of Statistics and Applied Mathematics, Anhui University of Finance and Economics [2]Institute for Information and System Science, Xi'an Jiaotong University

出　　处：《Science China(Information Sciences)》2014年第1期107-120,共14页中国科学（信息科学）（英文版）

基　　金：supported by National Basic Research Program of China(973)(Grant No.2013CB329404);National Natural Science Foundation of China(Grant No.11101327)

摘　　要：For data sets of arbitrary shapes and densities, the existing clusterings have much space to be improved to obtain better results. In this paper, clustering is considered as a cognitive problem, and cognitive features are of vital importance to clustering. In combination with psychological experiment, we propose three cognitive features of clustering and model them as a flexible similarity measurement. Meanwhile a new clustering framework is put forward to integrate the cognitive features by employing the similarity measurement. The two attractive advantages are its low complexity and fitness for various types of data sets, such as data sets of diferent shapes and densities. Some synthetic and real data sets are employed to exhibit the superiority of the new clustering algorithm.For data sets of arbitrary shapes and densities, the existing clusterings have much space to be improved to obtain better results. In this paper, clustering is considered as a cognitive problem, and cognitive features are of vital importance to clustering. In combination with psychological experiment, we propose three cognitive features of clustering and model them as a flexible similarity measurement. Meanwhile a new clustering framework is put forward to integrate the cognitive features by employing the similarity measurement. The two attractive advantages are its low complexity and fitness for various types of data sets, such as data sets of diferent shapes and densities. Some synthetic and real data sets are employed to exhibit the superiority of the new clustering algorithm.

关 键 词：CLUSTERING direction consistent structural nearest neighborhood afecting field MANIFOLD 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]