Article
Mathematics, Applied
Yifen Ke, Changfeng Ma, Zhigang Jia, Yajun Xie, Riwei Liao
Summary: A novel quasi non-negative quaternion matrix factorization (QNQMF) model is proposed to address the non-negativity dropout problem of quaternion models in color image processing. The quaternion projected gradient algorithm and the quaternion alternating direction method of multipliers are used to implement QNQMF by formulating it as non-convex constraint quaternion optimization problems. Experimental results show that encoding algorithms on quaternions outperform those on RGB channels in color image reconstruction and face recognition, especially when dealing with large facial expressions and shooting angle variations.
JOURNAL OF SCIENTIFIC COMPUTING
(2023)
Article
Automation & Control Systems
Zhiwei Xing, Meng Wen, Jigen Peng, Jinqian Feng
Summary: The paper introduces a novel discriminative semi-supervised NMF (DSSNMF) algorithm that effectively utilizes label information from a portion of the data, with empirical experiments demonstrating its effectiveness.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2021)
Article
Computer Science, Artificial Intelligence
Gal Gilad, Itay Sason, Roded Sharan
Summary: Non-negative matrix factorization (NMF) is a popular method used to find low rank approximations of matrices, especially in genomics for interpreting mutation data. A key challenge in using NMF is determining the number of components. A new method, CV2K, is proposed in this study to automatically select this number based on cross validation and parsimony considerations. Results show that CV2K leads to improved predictions compared to previous approaches, even those involving human assessment.
MACHINE LEARNING-SCIENCE AND TECHNOLOGY
(2021)
Article
Biochemistry & Molecular Biology
Junmin Zhao, Yuanyuan Ma, Lifang Liu
Summary: In this paper, a new algorithm LVSNMF is introduced which simultaneously utilizes Laplacian and Vicus matrices to capture both global and local structure patterns. Experimental results demonstrate its superior performance in biological data analysis.
FRONTIERS IN MOLECULAR BIOSCIENCES
(2021)
Article
Genetics & Heredity
Saurav Mallik, Anasua Sarkar, Sagnik Nath, Ujjwal Maulik, Supantha Das, Soumen Kumar Pati, Soumadip Ghosh, Zhongming Zhao
Summary: In this study, a novel framework called three-factor penalized non-negative matrix factorization-based multiple kernel learning with soft margin hinge loss (3PNMF-MKL) was proposed for multi-modal data integration and gene signature detection. The algorithm identified a 50-gene signature using an acute myeloid leukemia cancer dataset and achieved a high classification AUC score.
FRONTIERS IN GENETICS
(2023)
Article
Biochemical Research Methods
Peng Peng, Yipu Zhang, Yongfeng Ju, Kaiming Wang, Gang Li, Vince D. Calhoun, Yu-Ping Wang
Summary: In this study, a novel algorithm called GJNMFO is proposed to integrate SNP, fMRI and DNA methylation data to identify risk genes, epigenetic factors, and abnormal brain regions associated with schizophrenia. The algorithm effectively discards unimportant features and improves accuracy by introducing orthogonal constraints and group sparse processing.
IEEE-ACM TRANSACTIONS ON COMPUTATIONAL BIOLOGY AND BIOINFORMATICS
(2022)
Article
Computer Science, Information Systems
Diego Salazar, Juan Rios, Sara Aceros, Oscar Florez-Vargas, Carlos Valencia
Summary: The integration of data from different sources using NMF and jNMF methods can facilitate clustering and interpretation, but they may not effectively identify nonlinear patterns. A new variant called Kernel jNMF is proposed to address this limitation, showing better performance in clustering and interpretation.
Article
Automation & Control Systems
Imtiaz Ahmed, Xia Ben Hu, Mithun P. Acharya, Yu Ding
Summary: Dimensionality reduction is crucial for competitive performance in unsupervised learning, and Non-negative matrix factorization (NMF) is widely used for this purpose. However, in the presence of nonlinear manifold structure, NMF may not perform well. To address this, a neighborhood structure-assisted NMF method is proposed, showing superior performance through empirical comparisons and property analysis.
JOURNAL OF MACHINE LEARNING RESEARCH
(2021)
Article
Engineering, Mechanical
Mateusz Gabor, Rafal Zdunek, Radoslaw Zimroz, Jacek Wodecki, Agnieszka Wylomanska
Summary: In this study, a novel non-negative tensor factorization (NTF)-based method is proposed for vibration-based local damage detection in rolling element bearings. The time-frequency method is used to decompose the non-stationary diagnostic signal from faulty machines. Multi-linear NTF-based components are extracted from a 3D array of time-frequency representations of the observed signal, allowing for efficient separation of informative and non-informative components. Experiments on synthetic and real signals demonstrate the high efficiency of this method compared to the existing non-negative matrix factorization approach.
MECHANICAL SYSTEMS AND SIGNAL PROCESSING
(2023)
Article
Engineering, Electrical & Electronic
Xiangyu Liu, Peng Song
Summary: This paper proposes a novel multi-view clustering model called virtual label guided multi-view non-negative matrix factorization (VLMNMF). It utilizes virtual label information to guide the learning of latent representation and integrates the latent representation learning and clustering process into a joint framework. Experimental results demonstrate the effectiveness of the proposed method.
DIGITAL SIGNAL PROCESSING
(2023)
Article
Computer Science, Artificial Intelligence
Khanh Luong, Richi Nayak, Thirunavukarasu Balasubramaniam, Md Abul Bashar
Summary: This paper proposes a deep non-negative matrix factorization-based framework for effective multi-view data clustering by uncovering the non-linear relationships and intrinsic components of the data. The framework effectively incorporates the optimal manifold of multi-view data and outperforms existing multi-view matrix factorization-based methods.
PATTERN RECOGNITION
(2022)
Article
Computer Science, Artificial Intelligence
Yufu Chen, Zhiqi Lei, Yanghui Rao, Haoran Xie, Fu Lee Wang, Jian Yin, Qing Li
Summary: Non-negative Matrix Tri-Factorization (NMTF) is a novel paradigm for data mining and dimensionality reduction that has gained attention due to its notable performance and elegant mathematical derivation. However, existing NMTF-based methods suffer from high computational complexity. This paper proposes a parallel and scalable NMTF-based algorithm for text data co-clustering.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2023)
Article
Automation & Control Systems
Philip K. Hopke, Yunle Chen, David Q. Rich, Dennis Mooibroek, Uwayemi M. Sofowote
Summary: Over the past decade, Positive Matrix Factorization (PMF) has become the most commonly used tool for source apportionment of air pollutants. However, the implementation developed by the U.S. Environmental Protection Agency (EPA-PMF) has limitations in handling large data sets. In this study, we propose a protocol using the multilinear engine (ME-2) to analyze larger data sets and present results for particle size distributions in Rochester, New York.
CHEMOMETRICS AND INTELLIGENT LABORATORY SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Ghufran Ahmad Khan, Jie Hu, Tianrui Li, Bassoma Diallo, Hongjun Wang
Summary: This study introduces a new NMF clustering method for multi-view data, utilizing manifold regularization to preserve the local geometrical structure of the data space and achieve better clustering performance through iterative optimization strategy.
INTERNATIONAL JOURNAL OF MACHINE LEARNING AND CYBERNETICS
(2022)
Article
Biotechnology & Applied Microbiology
Ko Abe, Masaaki Hirayama, Kinji Ohno, Teppei Shimamura
Summary: The BALSAMICO hierarchical Bayesian framework accurately estimates parameters needed to analyze the connections between microbial community systems and their environments, and effectively detects these communities in real-world circumstances.