Cluster label aligning algorithm based on programming model
Haixiong Fan, Fuxian Liu, Lu Xia · 2012
Cluster ensembles can improve single clusters performance effectively, but the cluster labels are not fused directly due to lack of prior information supervising. Based on analysis and research on existing methods, label aligning was convert to the assign problem. And two cluster label aligning algorithms were proposed, which use overlap similarity rate as the coefficient matrix, Hungary and implicit enumeration algorithm as basic algorithms. Lastly, experimental simulation and capability analysis were completed, and results prove the availability and applicability of the new algorithms.