Article
Engineering, Electrical & Electronic
Xinglong Pei, Xiaoyang Zheng, Jinliang Wu
Summary: The paper introduces a novel Transformer convolution network (TCN) based on transfer learning, which has achieved highly accurate fault diagnosis. Experimental results demonstrate the robustness and effectiveness of the proposed method.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2021)
Article
Public, Environmental & Occupational Health
Sahar Arooj, Muhammad Atta-ur-Rahman, Muhammad Zubair, Muhammad Farhan Khan, Khalid Alissa, Muhammad Adnan Khan, Amir Mosavi
Summary: Breast cancer is a common type of cancer, and accurate identification and classification are crucial for early detection and treatment. This study proposes a model using transfer learning technique, trained on multiple datasets, achieving higher accuracy.
FRONTIERS IN PUBLIC HEALTH
(2022)
Article
Computer Science, Interdisciplinary Applications
Brijesh Kumar Chaurasia, Harsh Raj, Shreya Singh Rathour, Piyush Bhushan Singh
Summary: This study proposes an ensemble model driven by transfer learning for the detection of diabetic retinopathy (DR). DR is an eye problem caused by diabetes, where the retinal blood vessels deteriorate. By utilizing retinal scans and computer vision-based methods, the model is able to automatically identify the condition. Six deep learning-based convolutional neural network models were used, and a data-preprocessing strategy was applied to improve accuracy and reduce training costs. Experimental results show that the suggested model outperforms existing approaches, achieving an accuracy of up to 98% and effectively detecting the stage of DR.
MEDICAL & BIOLOGICAL ENGINEERING & COMPUTING
(2023)
Article
Engineering, Electrical & Electronic
Chaoyi Zhu, Zhenghua Chen, Rui Zhao, Jinjiang Wang, Ruqiang Yan
Summary: The article discusses an important research direction in the field of machine health monitoring: Explainable MHMS (EMHMS). By introducing a specific convolutional neural network structure and gradient-based methods to explain the model's decisions, it aims to balance the precision and explainability of predictive models.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2021)
Article
Computer Science, Interdisciplinary Applications
Bufan Liu, Yingfeng Zhang, Jingxiang Lv, Arfan Majeed, Chun-Hsien Chen, Dang Zhang
Summary: The rapid development of the industrial Internet of Things has promoted manufacturing towards the cyber-physical system, where highly accurate process recognition is crucial for proactive monitoring. This study proposes a deep transfer learning-based manufacturing process recognition approach, which achieves better accuracy with less training time and fewer training samples.
ROBOTICS AND COMPUTER-INTEGRATED MANUFACTURING
(2021)
Article
Computer Science, Information Systems
Lipeng Zhu, Junjie Lin
Summary: This article introduces a novel spatiotemporal correlation learning scheme (SCLS) for online missing PMU data correction (MPDC) in challenging measurement contexts, showing high efficacy in refining correction results and filtering out potential noises.
IEEE INTERNET OF THINGS JOURNAL
(2021)
Article
Chemistry, Analytical
Lauren J. Wong, Alan J. Michaels
Summary: Transfer learning is widely used in computer vision and natural language processing fields, but its application in radio frequency machine learning is limited. This paper proposes a tailored taxonomy for radio frequency applications and outlines directions for future research.
Article
Computer Science, Information Systems
Abdessalam Hattab, Ali Behloul
Summary: With the increasing demand for user recognition, experts recommend the use of biometric identification technology in application development. Single biometric modalities are insufficient for high-security requirements, leading to the rise of multimodal systems. This research proposes a robust multimodal biometric recognition system based on face and iris fusion, achieving exceptional performance and reliability.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Engineering, Electrical & Electronic
Brandon Hobley, Michal Mackiewicz, Julie Bremner, Tony Dolphin, Riccardo Arosio
Summary: This article assessed the reliability of crowdsourced labels for estuarine vegetation and unvegetated sediment, and found that the accuracy of the labels was influenced by the expertise and familiarity of the participants. The results also confirmed that biases in participant annotation were propagated in the performance of the deep learning models. Additionally, it was shown that combining in situ and crowdsourced labels improved the performance of the models compared to using only in situ labels.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2023)
Article
Radiology, Nuclear Medicine & Medical Imaging
Lin Li, Xiaohui Lin, Tingting Liao, Rushan Ouyang, Meng Li, Jialin Yuan, Jie Ma
Summary: This study demonstrated the clinical application potential of a convolutional neural network (CNN)-based deep learning system in breast cancer screening and diagnosis in Asian women. The DL system showed higher sensitivity in mass detection compared to junior radiologists and was not affected by breast density.
QUANTITATIVE IMAGING IN MEDICINE AND SURGERY
(2023)
Article
Computer Science, Information Systems
Shivani Gaba, Ishan Budhiraja, Vimal Kumar, Sahil Garg, Georges Kaddoum, Mohammad Mehedi Hassan
Summary: This paper provides a detailed review of various deep learning architectures and models, with a focus on a specific convolutional neural network model. It discusses the working principles of convolutional neural networks and their components and presents various models from LeNet to AlexNet, GoogleNet, VGGNet, ResNet, DenseNet, Xception, PNAS/ENAS, and EfficientNet. The challenges associated with different network architectures are also summarized. The paper concludes with a discussion of the frameworks, datasets, applications, and accuracy of each model, serving as a future scope in the field.
COMPUTER COMMUNICATIONS
(2022)
Article
Computer Science, Information Systems
Neha Thakur, Pardeep Kumar, Amit Kumar
Summary: This paper examines and evaluates the current machine and deep learning techniques for breast cancer identification and classification based on breast screening images. Ten research questions are formulated to determine the scope of the review, and an extensive review of research papers and book chapters from 2010 to 2021 is conducted. The review identifies several issues, including image modalities, segmentation techniques, features, and evaluation metrics. The findings show that digital mammograms are commonly used for breast cancer identification, and support vector machine and convolutional neural network are the most frequently used classifiers. Deep learning techniques have proven effective in image analysis.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Engineering, Electrical & Electronic
Jian Duan, Cheng Hu, Hongdi Zhou, Xiaobin Zhan, Feng Xiong, Tielin Shi
Summary: A novel deep learning framework named ECA-SDResNet-SVM is proposed for bearing health condition recognition. The framework utilizes Siamese neural network to learn features and introduces ECA module to highlight sensitive components, and then uses SVM model to identify bearing fault status. Experimental results show that the framework outperforms other models in terms of sample scales and noise levels, and exhibits prominent performance.
IEEE SENSORS JOURNAL
(2023)
Article
Computer Science, Information Systems
T. Muthamilselvan, K. Brindha, Sudha Senthilkumar, Saransh, Jyotir Moy Chatterjee, Yu-Chen Hu
Summary: This research proposes a method for detecting human emotions through facial expressions using computer vision and machine learning. The method consists of three phases: background noise removal using a convolutional neural network, facial feature extraction, and feature selection using binary whale optimization. Experimental results showed accuracy rates of 98.35%, 99.42%, 96.6%, and 64.98% on different datasets.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Computer Science, Information Systems
Samah A. Gamel, Esraa Hassan, Nora El-Rashidy, Fatma M. Talaat
Summary: The COVID-19 pandemic has had a significant impact on human migration worldwide, affecting transportation patterns in cities. This paper analyzes the relationship between COVID-19 and transportation using correlations and machine learning techniques, and introduces a Traffic Prediction Module (TPM) to predict the impact of COVID-19 on transportation. The results indicate a strong correlation between the spread of COVID-19 and transportation patterns, and the CNN has a high accuracy rate in predicting these impacts.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Hongbing Shang, Chuang Sun, Jinxin Liu, Xuefeng Chen, Ruqiang Yan
Summary: Surface defect detection plays a crucial role in intelligent manufacturing and product life-cycle management. Existing methods mainly rely on convolutional architectures, but the limited receptive field poses challenges for performance improvement. Transformer-based models, with their ability to model long-range dependencies, have achieved success in computer vision. However, using Transformer models without modification lacks defect awareness and relevance.
ADVANCED ENGINEERING INFORMATICS
(2023)
Article
Computer Science, Interdisciplinary Applications
Ruochen Jin, Laihao Yang, Zhibo Yang, Yu Sun, Zhu Mao, Ruqiang Yan, Xuefeng Chen
Summary: Digital twin has great potential for health monitoring of aero-engine blade. However, the lack of accurate connection method between the digital-twin model and the physical blade is a challenging issue. Blade tip timing (BTT) is an effective non-contact measurement method but the signal is generally incomplete and under-sampled. A novel method based on atomic norm soft thresholding (AST) is proposed for blade vibration parameter reconstruction from the undersampled BTT signal, providing accurate blade vibration information for the digital-twin model.
STRUCTURAL AND MULTIDISCIPLINARY OPTIMIZATION
(2023)
Article
Automation & Control Systems
Chenye Hu, Jingyao Wu, Chuang Sun, Ruqiang Yan, Xuefeng Chen
Summary: Recent research has made significant progress in intelligent fault diagnosis algorithms. However, the lack of labeled data in practical scenarios poses challenges for data annotation and increases the risk of overfitting, hindering industrial applications. To address this, this article proposes a framework that combines self-supervised learning on unlabeled data with supervised learning on few labeled data to enhance the learnable data's capacity. The framework includes a time-amplitude signal augmentation technique, interinstance transform-consistency learning for domain-invariant features, intratemporal relation matching to improve temporal discriminability, and an uncertainty-based dynamic weighting mechanism for multitask optimization stability. Experimental results on both open-source and self-designed datasets demonstrate the superiority of the proposed framework over other supervised and semi-supervised methods.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2023)
Article
Automation & Control Systems
Zengkun Wang, Zhibo Yang, Guangrong Teng, Ruqiang Yan, Shaohua Tian, Haoqi Li, Jiahui Cao, Xuefeng Chen
Summary: MUSIC is widely used for frequency estimation but cannot estimate the amplitude. This article rederives MUSIC based on real signals and proposes an amplitude-identifiable MUSIC (Aid-MUSIC) approach to recover the amplitude information. Simulations and experiments demonstrate that Aid-MUSIC can simultaneously and stably extract the amplitude and frequency of asynchronous frequency components.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2023)
Article
Engineering, Mechanical
Zuogang Shang, Zhibin Zhao, Ruqiang Yan, Xuefeng Chen
Summary: Machine anomaly detection is the task of detecting abnormal machine conditions using collected monitoring data. Autoencoder (AE) based unsupervised anomaly detection (UAD) has gained increasing attention for mechanical equipment. However, the raw monitoring data may be polluted by abnormal data, and without effective regularization, AE-based methods would overfit these polluted data. To address this issue, the core loss is designed to perform AE-based UAD in a model-agnostic and end-to-end manner under data pollution.
MECHANICAL SYSTEMS AND SIGNAL PROCESSING
(2023)
Article
Mechanics
Xingyu Wang, Yafei Xu, Yuqing Cui, Wenkang Li, Liuyang Zhang, Ruqiang Yan, Xuefeng Chen
Summary: With the widespread use of GFRP composites in engineering structures, the quality inspection of GFRPs is urgently needed. The combination of terahertz spectroscopy and artificial intelligence has shown great potential for automatic defect identification inside composites. In this study, a deformable attention convolutional neural network (DACNN) framework-based THz characterization system is proposed to automatically locate and image defects. Experimental results validate the effectiveness of the proposed system, providing a new solution for intelligent and automatic THz characterization of internal debonding defects of composites.
COMPOSITE STRUCTURES
(2023)
Article
Automation & Control Systems
Chenye Hu, Jingyao Wu, Chuang Sun, Xuefeng Chen, Ruqiang Yan
Summary: Flight regime recognition plays a critical role in ensuring flight safety and maintenance decisions. Existing methods suffer from imprecise boundary localization and low recognition accuracy. In this study, a new intelligent temporal detection network is proposed to handle long flight sequences and generate multiple regime boxes. Specific model structures are designed for flight data, including adaptive graph embedding, multi-scale Transformer encoder, and balanced joint loss. Experimental results demonstrate the effectiveness of the proposed method in accurate and boundary-sensitive regime recognition.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2023)
Article
Engineering, Electrical & Electronic
Yuansheng Cheng, Zhixiong Li, Grzegorz M. Krolczyk, Chunsheng Yang, Ruqiang Yan
Summary: Vibration measurement is crucial for monitoring and controlling vibrations. Noncontact vibration measurement is more practical than contact methods, but noncontact methods often require a high standard of the light field environment. To solve this problem, a new binocular vision method is proposed, which allows noncontact vibration measurement in different light fields. This method first recognizes multitarget objects in the same image using a YOLOv5 model, generating bounding boxes. A depth image is generated through binocular vision and kept synchronized with the target image. An optimal depth value decision algorithm is developed to determine the 3-D real-time coordinates of each object based on the bounding box and its corresponding depth image. As a result, vibrations of multitarget objects can be measured simultaneously. Experimental tests show accurate results and the ability of the proposed noncontact method to detect very low-frequency vibrations.
IEEE SENSORS JOURNAL
(2023)
Article
Automation & Control Systems
Hongbing Shang, Jingyao Wu, Chuang Sun, Jinxin Liu, Xuefeng Chen, Ruqiang Yan
Summary: This article proposes an intelligent borescope inspection method to detect surface damage on aeroengine blades. The method efficiently models pixel-to-pixel relations and handles weak damage information caused by background noise and unsatisfactory illumination. It also incorporates label relations as prior knowledge and fuses image features and label features for mode recognition and damage localization.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2023)
Article
Automation & Control Systems
Yongyi Chen, Dan Zhang, Kunpeng Zhu, Ruqiang Yan
Summary: In this article, a new activation function called Parameter-free Adaptively Swish (PASwish) is developed to improve the flexibility and generalization ability of deep learning frameworks in industrial scenarios with changing operating conditions. Additionally, deep parameter-free cosine networks with PASwish are proposed to adjust network weights based on domain-specific and domain-invariant features. The proposed method achieves better performance in cross-domain fault diagnosis compared to current studies, with an average accuracy of 95.16% (+/- 1.76%) on 72 transfer tasks.
IEEE-ASME TRANSACTIONS ON MECHATRONICS
(2023)
Article
Computer Science, Artificial Intelligence
Yadong Xu, Ke Feng, Xiaoan Yan, Ruqiang Yan, Qing Ni, Beibei Sun, Zihao Lei, Yongchao Zhang, Zheng Liu
Summary: This paper introduces a novel convolutional fusion framework called CFCNN, which extracts multilevel modality-specific features from different mechanical signals using the multiscale shrinkage denoising module (MSDM), and explores the intrinsic correlations and integrates cross-modal features through the central fusion module (CFM). Moreover, an online label smoothing training (OLST) strategy is applied to improve the classification performance of CFCNN. The efficacy of the developed CFCNN is verified using the cylindrical rolling bearing dataset and the planetary gearbox dataset.
INFORMATION FUSION
(2023)
Article
Engineering, Industrial
Sinan Li, Tianfu Li, Chuang Sun, Ruqiang Yan, Xuefeng Chen
Summary: This article proposes a novel Multilayer Grad-CAM (MLG-CAM) as a tool to explain deep neural networks, improving network explainability. Three indicators are defined to quantify the explainability of deep neural networks. Experimental results demonstrate that MLG-CAM not only highlights cyclostationary impulses in the time domain but also emphasizes fault characteristic frequencies in the frequency domain. These results indicate that MLG-CAM is an effective way to explain deep neural networks and build trust in networks.
JOURNAL OF MANUFACTURING SYSTEMS
(2023)
Article
Engineering, Mechanical
Qisheng Wang, Xin Lin, Xianyin Duan, Ruqiang Yan, Jerry Ying Hsi Fuh, Kunpeng Zhu
Summary: Laser powder bed fusion (L-PBF) is a metal additive manufacturing process with potential for high-performance metal components. However, stability and repeatability issues limit its industrial application. To ensure product quality, process monitoring and control are crucial. A new motion feature is extracted and a classification model is constructed to identify the changing melt pool states during the L-PBF process. The Gaussian process classification (GPC) model achieves better recognition results compared to other models, with an overall recognition rate of 87.1%.
MECHANICAL SYSTEMS AND SIGNAL PROCESSING
(2023)
Article
Computer Science, Hardware & Architecture
Wenjun Sun, Ruqiang Yan, Ruibing Jin, Jiawen Xu, Yuan Yang, Zhenghua Chen
Summary: This article addresses the limitations of the Transformer model in fault diagnosis and proposes two improvements: replacing or surpassing the attention module with convolutional layers to reduce computation cost, and developing a lightweight Transformer called LiteFormer using depth-wise convolutional layers. Extensive experiments demonstrate that these improvements significantly enhance the fault classification accuracy and computation efficiency of the model.
IEEE TRANSACTIONS ON RELIABILITY
(2023)