Abstract:As a popular data representation technique, Nonnegative matrix factorization (NMF) has been widely applied in edge computing, information retrieval and pattern recognition. Although it can learn parts-based data representations, existing NMF-based algorithms fail to integrate local and global structures of data to steer matrix factorization. Meanwhile, semi-supervised ones ignore the important role of instances from different classes in learning the representation. To solve such an issue, we propose a novel semi-supervised NMF approach via joint graph regularization and constraint propagation for edge computing, called robust constrained nonnegative matrix factorization (RCNMF), which learns robust discriminative representations by leveraging the power of both L2,1-norm NMF and constraint propagation. Specifically, RCNMF explicitly exploits global and local structures of data to make latent representations of instances involved by the same class closer and those of instances involved by different classes farther. Furthermore, RCNMF introduces the L2,1-norm cost function for addressing the problems of noise and outliers. Moreover, L2,1-norm constraints on the factorial matrix are used to ensure the new representation sparse in rows. Finally, we exploit an optimization algorithm to solve the proposed framework. The convergence of such an optimization algorithm has been proven theoretically and empirically. Empirical experiments show that the proposed RCNMF is superior to other state-of-the-art algorithms.

Correntropy induced metric based graph regularized non-negative matrix factorization

Correntropy Supervised Non-Negative Matrix Factorization

Data representation using robust nonnegative matrix factorization for edge computing

Graph regularized non-negative matrix factorization with sparse coding

Adaptive graph regularized nonnegative matrix factorization for data representation

Convergence Analysis of Graph Regularized Non-Negative Matrix Factorization

Feature Extraction Via Multi-View Non-Negative Matrix Factorization with Local Graph Regularization.

Robust non-negative matrix factorization with multiple correntropy-induced hypergraph regularizer

Spatially correlated nonnegative matrix factorization

Graph Regularized Nonnegative Matrix Factorization with Sparse Coding

Spatially correlated nonnegative matrix factorization for image analysis

Robust Structured Nonnegative Matrix Factorization for Image Representation.

Image Feature Extraction via Graph Embedding Regularized Projective Non-negative Matrix Factorization.

Semi-supervised Convex Nonnegative Matrix Factorizations with Graph Regularized for Image Representation

Non-negative Matrix Factorization on Manifold

Incremental nonnegative matrix factorization based on correlation and graph regularization for matrix completion

A locally weighted sparse graph regularized Non-Negative Matrix Factorization method.

Constrained Nonnegative Matrix Factorization for Image Representation

Graph Regularized Sparse Non-Negative Matrix Factorization for Clustering

Non-negative Local Coordinate Factorization for Image Representation.

Truncated Cauchy Non-Negative Matrix Factorization.