Clustering high-dimensional data and making sense out of its result is a challenging problem. In this paper, we present a weakly supervised nonnegative matrix factorization (NMF) and its symmetric version that take into account various prior information via regularization in clustering applications. Unlike many other existing methods, the proposed weakly supervised NMF methods provide interpretable and flexible outputs by directly incorporating various forms of prior information. Furthermore, the proposed methods maintain a comparable computational complexity to the standard NMF under an alternating nonnegativity-constrained least squares framework. By using real-world data, we conduct quantitative analyses to compare our methods against other semi-supervised clustering methods. We also present the use cases where the proposed methods lead to semantically meaningful and accurate clustering results by properly utilizing user-driven prior information.
- Nonnegative matrix factorization
- Semi-supervised clustering
- User-driven clustering
ASJC Scopus subject areas
- Information Systems
- Computer Science Applications
- Computer Networks and Communications