Article
Automation & Control Systems
Yongqing Wang, Bo Qin, Kuo Liu, Mingrui Shen, Mengmeng Niu, Lingsheng Han
Summary: The article introduces a multitask learning method based on deep belief networks for predicting tool wear and part surface quality. Experimental results show that the proposed method can improve prediction accuracy and reduce computing time.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2021)
Article
Computer Science, Artificial Intelligence
Ao Li, Cong Feng, Yuan Cheng, Yingtao Zhang, Hailu Yang
Summary: This paper proposes a new method for incomplete multiview subspace clustering. By using multiple kernel completion, low-redundant representation learning, and weighted tensor low-rank constraint, intact and compact subspaces can be obtained. Instead of the traditional pairwise subspace fusion, the proposed method fuses multiview subspaces with a weighted tensor low-rank constraint, which explores higher-order relationships among views and assigns appropriate weights to each view. Extensive experiments demonstrate the effectiveness of the proposed method.
INFORMATION FUSION
(2024)
Article
Computer Science, Artificial Intelligence
Wei Chang, Feiping Nie, Yijie Zhi, Rong Wang, Xuelong Li
Summary: Multitask learning is a joint learning paradigm that combines multiple related tasks to improve performance. It has been observed that different tasks share a low-dimensional common subspace. To approximate the rank minimization problem, two regularization-based models are proposed to minimize the k minimal singular values. Compared to the standard trace norm, these models provide tighter approximations and better capture the low-dimensional subspace among multiple tasks.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Engineering, Civil
Dan Luo, Dong Zhao, Zijian Cao, Mingyao Wu, Liang Liu, Huadong Ma
Summary: Traffic prediction under different conditions is important but challenging. This study proposes a novel deep learning model called M(3)AN to effectively handle traffic prediction under abnormal conditions, and achieves better results compared to existing methods on two real-world traffic datasets.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2023)
Article
Mathematics
Cosimo Flavi
Summary: We determine the border rank of each power of any quadratic form in three variables. We primarily focus on non-degenerate quadratic forms by considering the apolar ideal of each power. Using a specific ideal contained in the apolar ideal for each power of a quadratic form in three variables, we establish the border rank of any power to be equal to the rank of its middle catalecticant matrix.
JOURNAL OF ALGEBRA
(2023)
Article
Environmental Sciences
Le Dong, Yuan Yuan
Summary: This paper introduces a novel method for hyperspectral image unmixing based on non-negative tensor factorization, which uses different low-rank representation modes to explore spectral, spatial, and non-local similarity. By dividing HSI into blocks, designing low-rank regularization, and integrating the sparsity of the abundance tensor, the unmixing performance is improved.
Article
Computer Science, Artificial Intelligence
Senhong Wang, Jiangzhong Cao, Fangyuan Lei, Jianjian Jiang, Qingyun Dai, Bingo Wing-Kuen Ling
Summary: This paper presents a novel and flexible unified graph learning framework for handling incomplete multi-view data. By improving the anchor selection strategy and introducing a cross-view anchor graph fusion mechanism, it effectively captures the intra-view and inter-view nonlinear relations. Experimental results demonstrate that the proposed method outperforms other methods in terms of clustering ability and time-consuming.
APPLIED INTELLIGENCE
(2023)
Review
Green & Sustainable Science & Technology
Xuming Zeng, Zinan Wang, Hao Wang, Shengyan Zhu, Shaofeng Chen
Summary: The condition of drainage pipes has a significant impact on the urban environment and human health. However, due to the location and structure of drainage pipes, conducting cost-effective pipeline investigation and evaluation is challenging. This study synthesizes and analyzes the four most commonly used drainage pipeline evaluation standards to summarize the deterioration and breakage patterns of drainage pipes. By integrating literature and engineering experience, the common pipe breakage patterns are also summarized. A comprehensive system of influencing factors for the condition of drainage pipes, including physical, environmental, and operational factors, is established, and the mechanisms of each influencing factor are summarized. Additionally, this study categorizes physical, statistical, and AI models and their representative models, and reviews the research progress of current mainstream drainage-pipe deterioration and breakage prediction models in terms of their principles and application progress.
Article
Computer Science, Information Systems
Yangfan Du, Gui-Fu Lu, Guangyan Ji
Summary: In this paper, a novel multi-view clustering method RONGL/MVC is proposed to solve the problem of multi-view clustering. The method constructs initial graphs and optimizes them to improve the clustering performance. Experimental results show that RONGL/MVC outperforms existing methods.
INFORMATION SCIENCES
(2023)
Article
Humanities, Multidisciplinary
Rebecca Piovesan, Elena Tesser, Lara Maritan, Gloria Zaccariello, Claudio Mazzoli, Fabrizio Antonelli
Summary: The HYPERION EU project aims to develop a Decision Support System for the sustainable reconstruction and resilience of historic areas facing climate change and extreme events. The study of the ornamental stones and deterioration patterns of the Clock Tower in Venice provides valuable insights into the decay of stone artifacts. The most common forms of deterioration include black crusts, patinas, discoloration, and erosion-related patterns.
Article
Computer Science, Artificial Intelligence
Tae San Kim, So Young Sohn
Summary: The research proposes a multi-task learning method based on convolution neural networks to better reflect the relationship between remaining useful life estimation and health status detection process, and it shows superior performance to existing baseline models in experiments using the C-MAPSS dataset for aero-engine unit prognostics.
JOURNAL OF INTELLIGENT MANUFACTURING
(2021)
Article
Computer Science, Interdisciplinary Applications
Yue Han, Qiu-Hua Lin, Li-Dan Kuang, Xiao-Feng Gong, Fengyu Cong, Yu-Ping Wang, Vince D. Calhoun
Summary: Tucker decomposition is commonly used for analyzing multi-subject fMRI data, but traditional methods are insufficient for extracting common patterns across subjects. In this study, we propose a low rank Tucker-2 model with spatial sparsity constraint to analyze multi-subject fMRI data. The results demonstrate that this method is more effective in extracting common spatial and temporal components compared to other algorithms, and the features extracted from the core tensor show promise for subject classification.
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2022)
Article
Computer Science, Information Systems
Weize Sun, Peng Zhang, Jingxin Xu, Huochao Tan
Summary: This paper proposes two algorithms for tensor completion, which are verified through comparisons of synthesis data and real world data to have advantages in recovery accuracy and computational complexity.
CMC-COMPUTERS MATERIALS & CONTINUA
(2021)
Article
Computer Science, Artificial Intelligence
Guoqing Liu, Hongwei Ge, Ting Li, Shuzhi Su, Shuangxi Wang
Summary: In this paper, a multi-view subspace enhanced representation of manifold regularization and low-rank tensor constraint (MSERMLRT) method is proposed to extract manifold information from multi-view data and improve clustering performance. A tensor is utilized to explore correlations between views and reduce redundant information. The model also incorporates manifold information and enforces a sparse constraint to enhance the diagonal block structure of the subspace representation, improving clustering effects. The effectiveness of the MSERMLRT model is demonstrated through experiments on challenging datasets.
INTERNATIONAL JOURNAL OF MACHINE LEARNING AND CYBERNETICS
(2023)
Article
Computer Science, Artificial Intelligence
Guangyan Ji, Gui-Fu Lu
Summary: Traditional multi-view clustering (MVC) cannot handle the problem of incomplete views, which often occurs in real-life scenarios. This paper proposes a novel method, CLIMVC/LTC, which combines missing view completion, low-rank tensor constraint, and consensus representation learning to address this issue. CLIMVC/LTC achieves improved clustering performance by jointly exploring higher-order correlations, cluster structure, and consistency between different views. Experimental results on well-known datasets demonstrate the effectiveness of CLIMVC/LTC.
INTERNATIONAL JOURNAL OF MACHINE LEARNING AND CYBERNETICS
(2023)
Article
Engineering, Industrial
Xiaoliang Yan, Reed Williams, Elena Arvanitis, Shreyes Melkote
Summary: This paper extends prior work by developing a semantic segmentation approach for machinable volume decomposition using pre-trained generative process capability models, providing manufacturability feedback and labels of candidate machining operations for query 3D parts.
JOURNAL OF MANUFACTURING SYSTEMS
(2024)
Article
Engineering, Industrial
Jing Huang, Zhifen Zhang, Rui Qin, Yanlong Yu, Guangrui Wen, Wei Cheng, Xuefeng Chen
Summary: In this study, a deep learning framework that combines interpretability and feature fusion is proposed for real-time monitoring of pipeline leaks. The proposed method extracts abstract feature details of leak acoustic emission signals through multi-level dynamic receptive fields and optimizes the learning process of the network using a feature fusion module. Experimental results show that the proposed method can effectively extract distinguishing features of leak acoustic emission signals, achieving higher recognition accuracy compared to typical deep learning methods. Additionally, feature map visualization demonstrates the physical interpretability of the proposed method in abstract feature extraction.
JOURNAL OF MANUFACTURING SYSTEMS
(2024)
