Article
Health Care Sciences & Services
Sehwan Moon, Hyunju Lee
Summary: The study introduces a joint deep semi-non-negative matrix factorization (JDSNMF) model, employing a hierarchical non-linear feature extraction approach to capture shared latent features from complex multi-omics data. The extracted latent features from JDSNMF can be used for various downstream tasks, including disease prediction and module analysis.
JOURNAL OF PERSONALIZED MEDICINE
(2021)
Article
Computer Science, Artificial Intelligence
Hongjie Zhang, Siyu Zhao, Wenwen Qiang, Yingyi Chen, Ling Jing
Summary: This study proposes a unified feature extraction framework based on contrastive learning with adaptive positive and negative samples (CL-FEFA) for unsupervised, supervised, and semi-supervised feature extraction. The framework constructs adaptive positive and negative samples to extract discriminative features, resulting in more compact intra-class embedded samples and more dispersed inter-class embedded samples. Numerical experiments demonstrate the significant advantages of the proposed framework in feature extraction.
Article
Computer Science, Artificial Intelligence
Youwei Wang, Lizhou Feng, Jianming Zhu, Yang Li, Fu Chen
Summary: An improved AdaBoost algorithm based on MOFS and WNMF is proposed in this paper to enhance classification performance effectively. Numerical experiments demonstrate that the proposed method achieves higher accuracy compared to traditional methods.
Article
Chemistry, Multidisciplinary
Seokjin Lee, Minhan Kim, Seunghyeon Shin, Sooyoung Park, Youngho Jeong
Summary: This paper develops feature extraction methods for weakly supervised sound event detection based on the NMF algorithm, achieving improved performance compared to other commonly used methods by training the frequency basis matrix and developing a non-iterative version.
APPLIED SCIENCES-BASEL
(2021)
Article
Computer Science, Artificial Intelligence
Xiangguang Dai, Keke Zhang, Juntang Li, Jiang Xiong, Nian Zhang, Huaqing Li
Summary: This paper proposes a robust semi-supervised non-negative matrix factorization method, RSNMF, for image clustering. By adding a weighted constraint on the noise matrix and incorporating manifold learning, better clustering performance on datasets contaminated by outliers and noise can be achieved. Additionally, utilizing discrete hashing learning method to constrain the learned subspace leads to a binary subspace from the original data.
COMPLEX & INTELLIGENT SYSTEMS
(2022)
Article
Automation & Control Systems
Chengcai Leng, Hai Zhang, Guorong Cai, Zhen Chen, Anup Basu
Summary: This paper presents a novel medical image registration algorithm named TV-GNMF, which utilizes non-negative matrix factorization and graph regularization, with total variation and graph regularization incorporated into NMF for denoising and enhancing discrimination power. The experiments show that the proposed TV-GNMF method outperforms other state-of-the-art algorithms.
IEEE-CAA JOURNAL OF AUTOMATICA SINICA
(2021)
Article
Biochemical Research Methods
Adel Mehrpooya, Farid Saberi-Movahed, Najmeh Azizizadeh, Mohammad Rezaei-Ravari, Farshad Saberi-Movahed, Mahdi Eftekhari, Iman Tavassoly
Summary: Extracting predictive features from complex high-dimensional multi-omic data is essential in systems pharmacology. This paper proposes three novel feature selection methods based on matrix factorization, which outperform other methods and successfully identify predictive features.
BRIEFINGS IN BIOINFORMATICS
(2022)
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
Naiyao Liang, Zuyuan Yang, Zhenni Li, Shengli Xie
Summary: Semi-supervised multi-view classification leverages information from labeled and unlabeled data to improve performance. A novel model is proposed in this paper to capture latent label information from unlabeled data and integrate it with multi-view representation learning, boosting each other.
Article
Computer Science, Artificial Intelligence
Xiaoxia Zhang, Xianjun Zhou, Lu Chen, Yanjun Liu
Summary: This paper proposes a Semi-Disentangled Non-negative Matrix Factorization (SDNMF) method, which separates the latent embeddings of users and items to improve the interpretability and generalizability of rating prediction methods.
APPLIED SOFT COMPUTING
(2023)
Article
Computer Science, Information Systems
Jeongyoung Hwang, Hyunju Lee
Summary: Integration of multiple biological datasets is important for understanding comprehensive biological mechanisms. This study presents a method called multimodal multitask matrix factorization (MMMF) to address the challenges of heterogeneity and imbalance in the data. Results show that MMMF outperforms other biomedical classification models and can be used for feature selection.
Article
Environmental Sciences
Md Touhid Islam, Md Rashedul Islam, Md Palash Uddin, Anwaar Ulhaq
Summary: Object classification in hyperspectral images is challenging due to high dimensionality and class imbalance. We propose a framework that addresses these challenges through dimensionality reduction and re-sampling. Our framework employs a subgroup-based dimensionality reduction technique and achieves class balance. The reduced and balanced data are processed by a hybrid CNN model to extract spectral-spatial features and improve classification accuracy.
Article
Computer Science, Artificial Intelligence
Zhiyuan Zou, Weibin Liu, Weiwei Xing
Summary: This paper proposes a novel fusion framework of adaptive nonnegative feature fusion for scene classification, which integrates nonnegative matrix factorization, adaptive feature fusion, and feature fusion boosting. Experiments demonstrate that the proposed method can efficiently handle multi-class scene problems and achieve remarkable classification performance.
PATTERN RECOGNITION
(2022)
Article
Computer Science, Artificial Intelligence
Zhili Zhao, Zhuoyue Gou, Yuhong Du, Jun Ma, Tongfeng Li, Ruisheng Zhang
Summary: ICP is a novel link prediction approach based on inductive matrix completion, which explores comprehensive node feature representation by combining different structural topology information with node importance properties. In twelve different real networks, ICP shows improved performance in terms of average AUC results, with significant improvements compared to the best baseline method.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Computer Science, Artificial Intelligence
Naiyao Liang, Zuyuan Yang, Zhenni Li, Shengli Xie, Weijun Sun
Summary: This study introduces a novel semi-supervised multi-view learning approach called Label Propagation based Non-negative Matrix Factorization (LPNMF) to address the problem of sparse labeled data. By constructing the intrinsic manifold structure of data and utilizing label propagation, the limited labeled data can be fully utilized. Experimental results demonstrate the advantages of this method over existing methods.
KNOWLEDGE-BASED SYSTEMS
(2021)