Fully Supervised Non-Negative Matrix Factorization for Feature Extraction

Austin, W. W.; Anderson, Dylan Zachary; Ghosh, Joydeep

doi:10.1109/igarss.2018.8518592

Cited by 3 publications

(2 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This can be done by minimizing a least squares error with a weighted penalty for the 1 -norm of the parameters. Prior authors have combined NMF with a linear regression procedure to maximize the predictive power of a classifier [10][11][12]. This is accomplished through a penalty function that combines NMF with another objective function-a (semi) supervised approach.…”

Section: Relation To Current Work and Contributionsmentioning

confidence: 99%

Continuous Semi-Supervised Nonnegative Matrix Factorization

et al. 2023

View full text Add to dashboard Cite

Nonnegative matrix factorization can be used to automatically detect topics within a corpus in an unsupervised fashion. The technique amounts to an approximation of a nonnegative matrix as the product of two nonnegative matrices of lower rank. In certain applications it is desirable to extract topics and use them to predict quantitative outcomes. In this paper, we show Nonnegative Matrix Factorization can be combined with regression on a continuous response variable by minimizing a penalty function that adds a weighted regression error to a matrix factorization error. We show theoretically that as the weighting increases, the regression error in training decreases weakly. We test our method on synthetic data and real data coming from Rate My Professors reviews to predict an instructor’s rating from the text in their reviews. In practice, when used as a dimensionality reduction method (when the number of topics chosen in the model is fewer than the true number of topics), the method performs better than doing regression after topics are identified—both during training and testing—and it retrains interpretability.

show abstract

Section: Relation To Current Work and Contributionsmentioning

confidence: 99%

Continuous Semi-Supervised Nonnegative Matrix Factorization

et al. 2023

View full text Add to dashboard Cite

show abstract

“…Therefore, the time complexity of the MADEL algorithm is O(tn). In contrast, the complexity of the traditional PCAbased NAD algorithm is O(tn 2 ) [35], and the tensor-based three-dimensional data anomaly detection and decomposition method's time complexity is approximated as O(kn 3 ) (k is a parameter) [23]. Moreover, the MADEL algorithm only uses two-dimensional data to reflect the spatial-temporal characteristic of the RTT values, which greatly reduces the space complexity.…”

Section: B Differential Decomposition Algorithmmentioning

confidence: 99%

Matrix Differential Decomposition-Based Anomaly Detection and Localization in NFV Networks

Chen

Wei³

et al. 2019

IEEE Access

View full text Add to dashboard Cite

Network function virtualization (NFV) is a promising network paradigm that enables the design and implementation of novel network services with lower cost, increased agility, and faster time-to-value. However, network anomalies caused by software malfunction, hardware failure, mis-configuration, or cyber attacks can greatly degrade the performance of NFV networks. A few matrix decomposition-based methods have shown their effectiveness in finding the existence of network-wide anomalies. However, a little attention has been paid to multiple anomalies detection and anomaly devices localization. To bridge this gap, in this paper, we propose a matrix differential decomposition (MDD)-based anomaly detection and localization algorithm for NFV networks. First, an NFV network prototype is built to investigate the property of NFV networks, and the effectiveness of traditional anomaly detection methods is evaluated. Second, we detail the MDD-based Anomaly DEtection and Localization (MADEL) algorithm. Finally, a series of experiments are conducted on three different NFV networks to evaluate the performance of the proposed algorithm. Experimental results show that the MADEL algorithm could effectively detect and localize different types of network anomalies.

show abstract