Constrained K-means with external information
Chen Zhigang, Li Xuan, Fan Yang · 2013
Constrained K-means clustering has been widely used in semi-supervised clustering. Background knowledge or priori information is usually presented as pair-wise constraints (must-link and cannot-link constraints) in the clustering objects. However, in many applications the relevant background knowledge about the data we want to analysis is not available. Instead we know there are some other relevant data which are not the same class as the clustering objects. We call these data as external information for the clustering objects and formalize it as a new constrained clustering problem with cannot-link constraints. Experiments on UCI datasets show that external information can effectively improve the clustering quality of clustering objects.