A deep discriminative and robust nonnegative matrix factorization network method with soft label constraint

Tang, Ming; Chen, Yiran; Zhao, Mengao; Bu, Haili; Xi, Shengnan

doi:10.1007/s00521-018-3554-6

Cited by 15 publications

(5 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Although the NMF algorithm and its variant algorithms have certain capabilities of feature extraction, the researchers believe that the underlying features of the sample cannot be obtained by shallow factorization [33]. Deep architectures have been widely applied particularly in image processing and analysis [23], [34], [38], since deep factorization provides high performance in data representation and efficient formulas behind deep learning [39]. However, there are still some challenges particularly in recognition [40].…”

Section: Nmfmentioning

confidence: 99%

“…The existing deep NMF algorithms are based on this multilayer factorization architecture [32], [33], [34]. The update formulas in each layer are…”

Section: Multilayer Non-negative Matrix Factorizationmentioning

confidence: 99%

“…Principal Component Analysis (PCA) [45], Kernel Principal Component Analysis (KPCA) [46], Direct Linear Discriminate Analysis (DLDA) [47], NMF [1], MNMF [29], PNMF [16], KNMF-RBF [17], SNMF [13], DSNMF [33] and DDRNMF [34] are chosen as comparison algorithms. In PCA, the eigenvector whose eigenvalue is large than 0.01 is kept.…”

Section: Databases and Experimental Parameters Settingmentioning

confidence: 99%

“…It abandons the idea of representing primitive samples based on the combination of basis images in classical NMF and weakens the semantic interpretability of the algorithm. In 2018, Tong et al proposed a deep discriminative and robust non-negative matrix factorization network method (DDRNMF) for obtaining differentiated, compact and robust data representations [34]. The algorithm achieves the data representation by minimizing an objective function with a soft label constraint.…”

Section: Introductionmentioning

confidence: 99%

“…Nowadays algorithms based on NMF and deep NMF focus on the coefficient matrix, including deep factorization on the coefficient matrix [33], [34] and adding different regularization constraints on the coefficient matrix [35], [36]. The optimization of the basis images matrix is indirect in these methods.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Deep Non-Negative Matrix Factorization Architecture Based on Underlying Basis Images Learning

Zhao

Wang

Pei

2021

IEEE Trans. Pattern Anal. Mach. Intell.

View full text Add to dashboard Cite

The non-negative matrix factorization (NMF) algorithm represents the original image as a linear combination of a set of basis images. This image representation method is in line with the idea of "parts constitute a whole" in human thinking. The existing deep NMF performs deep factorization on the coefficient matrix. In these methods, the basis images used to represent the original image is essentially obtained by factorizing the original images once. To extract features reflecting the deep localization characteristics of images, a novel deep NMF architecture based on underlying basis images learning is proposed for the first time. The architecture learns the underlying basis images by deep factorization on the basis images matrix. The deep factorization architecture proposed in this paper has strong interpretability. To implement this architecture, this paper proposes a deep non-negative basis matrix factorization algorithm to obtain the underlying basis images. Then, the objective function is established with an added regularization term, which directly constrains the basis images matrix to obtain the basis images with good local characteristics, and a regularized deep non-negative basis matrix factorization algorithm is proposed. The regularized deep nonlinear non-negative basis matrix factorization algorithm is also proposed to handle pattern recognition tasks with complex data. This paper also theoretically proves the convergence of the algorithm. Finally, the experimental results show that the deep NMF architecture based on the underlying basis images learning proposed in this paper can obtain better recognition performance than the other state-of-the-art methods.

show abstract