Semi-supervised pivotal-aware nonnegative matrix factorization with label and pairwise constraint propagation for data clustering

Xiaojun Yang, Tuoji Zhu, Siyuan Peng, Feiping Nie, Zhiping Lin

Abstract: Highlights•A novel and robust semi-supervised NMF method called SPNMF is proposed, which not only surmounts the limitations of prior semi-supervised NMF methods but also inherits their advantages, resulting in the extraction of a concise low-dimensional data representation.•The DCP algorithm is developed for SPNMF to enable the comprehensive and efficient propagation of limited label information throughout the entire dataset, manifesting as both pointwise and pairwise constraints.•The pivotal-aware technique is applied to improve the robustness of the proposed method.•The convergence, supervisory information utilization, and computational complexity of SPNMF are analyzed.

External IDs:dblp:journals/pr/YangZPNL25