Article
Engineering, Electrical & Electronic
Sudip Modak, Sayanjit Singha Roy, Rohit Bose, Soumya Chatterjee
Summary: In this study, a novel approach for automated detection and classification of focal EEG signals was proposed, utilizing cross wavelet transform and a customized CNN model. The experiment showed promising results, with 100% accuracy achieved for the delta rhythm and significantly reduced training time compared to existing CNN models.
IEEE SENSORS JOURNAL
(2021)
Article
Engineering, Multidisciplinary
Mantang Hu, Guofeng Wang, Kaile Ma, Zenghuan Cao, Shuai Yang
Summary: A method for bearing performance degradation assessment is proposed, using optimized empirical wavelet transform and fuzzy C-means model to improve the sensitivity and stability of the assessment method in extracting fault information.
Article
Computer Science, Information Systems
Xiangyu Zhao, Peng Huang, Xiangbo Shu
Summary: This paper investigates the issues in feature learning methods based on CNN and proposes a new module based on wavelet attention for image classification. Experimental results demonstrate significant improvements in accuracy using this approach.
MULTIMEDIA SYSTEMS
(2022)
Article
Engineering, Mechanical
Huan Wang, Zhiliang Liu, Dandan Peng, Ming J. Zuo
Summary: This paper proposes a multilayer wavelet attention convolutional neural network (MWA-CNN) for noise-robust machinery fault diagnosis. The framework aims to learn discriminative fault features from the wavelet domain, which allows the model to obtain better interpretability and superior performance than conventional time-domain-based CNNs. Experiments on high-speed aeronautical bearing and motor bearing datasets prove that the proposed method has excellent fault diagnosis ability and noise robustness, and the visual analysis of the attention mechanism contributes to the interpretability of CNN in the field of fault diagnosis.
MECHANICAL SYSTEMS AND SIGNAL PROCESSING
(2023)
Article
Engineering, Multidisciplinary
Subhashree Mohapatra, Girish Kumar Pati, Manohar Mishra, Tripti Swarnkar
Summary: This study proposes an intelligent method using empirical wavelet transform (EWT) and convolutional neural network (CNN) to classify alimentary canal diseases. The method achieves high accuracy and performance metrics in disease classification. A comparative study with other contemporary techniques is conducted to validate the efficacy of the proposed method.
AIN SHAMS ENGINEERING JOURNAL
(2023)
Article
Chemistry, Analytical
Muhammad Fayaz, Nurlan Torokeldiev, Samat Turdumamatov, Muhammad Shuaib Qureshi, Muhammad Bilal Qureshi, Jeonghwan Gwak
Summary: This paper introduces a brain MR image classification model based on discrete wavelet transform and convolutional neural network, which consists of three main stages: preprocessing, feature extraction, and classification. The proposed model has shown good results with high accuracy, outperforming other algorithms in comparison.
Article
Agronomy
Qiang Cui, Baohua Yang, Biyun Liu, Yunlong Li, Jingming Ning
Summary: This research proposes a method for tea recognition based on a lightweight convolutional neural network and support vector machine, utilizing wavelet feature figures. The results demonstrate that this method outperforms other techniques, achieving an accuracy rate of 98.7%. This study has important practical significance for the grading and quality assessment of tea.
Review
Agriculture, Multidisciplinary
Daoliang Li, Zhaoyang Song, Chaoqun Quan, Xianbao Xu, Chang Liu
Summary: This article discusses the importance of crop and livestock monitoring in agricultural production, as well as the application of image fusion technology in improving monitoring methods. It reviews the specific applications of image fusion in areas such as crop recognition, disease detection, and livestock health assessment, while also highlighting the challenges and future research directions in the field.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2021)
Article
Engineering, Marine
Bernice Kubicek, Ananya Sen Gupta, Ivars Kirsteins
Summary: This study trained convolutional neural networks on simulated data and related and explained the classifier results in the physical domain. The informative features used in discrimination were uncovered using the explainable artificial intelligence technique of gradient-weighted class activation mapping. It was found that the scalogram representation provided a negligible classification accuracy increase compared with the spectrograms, and the networks trained to discriminate between target geometries resulted in the highest accuracy, while the networks trained to discriminate the internal fluid of the target resulted in the lowest accuracy.
JOURNAL OF MARINE SCIENCE AND ENGINEERING
(2023)
Article
Engineering, Multidisciplinary
Hao-xiang He, Jia-cheng Zheng, Li-can Liao, Yan-jiang Chen
Summary: Traditional statistical pattern identification methods have limited ability to identify minor damage of bridges, but convolutional neural network combined with recurrence graph can achieve more accurate damage identification.
STRUCTURAL HEALTH MONITORING-AN INTERNATIONAL JOURNAL
(2021)
Article
Computer Science, Artificial Intelligence
Xishun Zhu, Zhengliang Lai, Yaru Liang, Jianping Xiong, Jianhua Wu
Summary: This paper proposes a generative image hiding algorithm based on residual convolutional neural network (ResCNN) in the wavelet domain to overcome the low hiding capacity and weak security measures of current hiding algorithms. The proposed algorithm effectively embeds high-frequency features of a secret image into a carrier image, leading to state-of-the-art results in terms of hiding capacity and security.
APPLIED SOFT COMPUTING
(2022)
Article
Automation & Control Systems
Yuxiang Wei, Huan Wang
Summary: This paper proposes a noise-robust framework for wafer defect recognition, which utilizes discrete wavelet transform for frequency learning. It introduces a learnable discrete wavelet transform layer and a frequency-location attention module. Experimental results show that the framework achieves excellent performance in detecting wafer defect images, with an accuracy of 98.84%, and outperforms other methods under high noise ratios.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2023)
Article
Engineering, Mechanical
Maohua Xiao, Wei Zhang, Kai Wen, Yue Zhu, Yilidaer Yiliyasi
Summary: In this study, Wavelet Packet Decomposition is used for feature extraction of vibration signals, which accurately distinguishes different states of the bearing through visualization of energy features using K-Means clustering. A fault diagnosis model based on BP Neural Network optimized by Beetle Algorithm is proposed to identify bearing faults with an accuracy exceeding 95% and certain anti-interference capability. Two experiments demonstrate the effectiveness of the model in fault diagnosis.
CHINESE JOURNAL OF MECHANICAL ENGINEERING
(2021)
Article
Chemistry, Analytical
Tabassum Islam Toma, Sunwoong Choi
Summary: This study proposes a novel deep learning model that combines recurrent neural networks and two-dimensional convolutional neural networks to improve the arrhythmia detection performance of imbalanced ECG signals. The experimental results show that the proposed model is very effective in arrhythmia detection and outperforms existing hierarchical network models.
Article
Computer Science, Information Systems
Wenbin Zou, Liang Chen, Yi Wu, Yunchen Zhang, Yuxiang Xu, Jun Shao
Summary: This article introduces a novel CNN-based super-resolution method named joint wavelet sub-bands guided network (JWSGN), which separates different frequency information of the image by the WT and recovers it through a multi-branch network. The method achieves better high-frequency detail reconstruction by using an edge extraction module and exploiting the complementary relationship between different frequencies.
IEEE TRANSACTIONS ON MULTIMEDIA
(2023)
Review
Computer Science, Artificial Intelligence
Feng He, Kai Bai, Yixin Zong, Yuan Zhou, Yimai Jing, Guoqiang Wu, Chen Wang
Summary: Makeup transfer (MT) is a technique that aims to transfer the makeup style from a given reference image to a source image while preserving face identity and background information. It has gained significant attention from scholars in recent years due to its wide range of applications and research value. Many methods based on Generative Adversarial Network (GAN) have been proposed, but there are still some challenges that need to be addressed in this field.
IET COMPUTER VISION
(2023)
Article
Computer Science, Artificial Intelligence
Guoqiang Wu, Feng He, Yuan Zhou, Yimai Jing, Xin Ning, Chen Wang, Bo Jin
Summary: This paper introduces an age-compensated makeup transformation framework based on homology continuity, aiming to solve the problem of unnatural face makeup images generated by existing methods due to the lack of consideration of age factor. By designing a new coding module, this method achieves stable and controllable age compensation effect.
IET COMPUTER VISION
(2023)
Article
Computer Science, Artificial Intelligence
Xin Ning, Weijuan Tian, Feng He, Xiao Bai, Le Sun, Weijun Li
Summary: In this study, a flexible and plastic hyper-sausage coverage function (HSCF) neuron model is proposed based on the mechanism of human cognition. A novel cross-entropy and volume-coverage (CE_VC) loss function is defined to enhance class separability and intra-class compactness. The introduced divisive iteration method adaptively determines the optimal number of HSCF neurons and a end-to-end learning framework is constructed. Experimental results demonstrate the effectiveness of the proposed method and its potential for boosting deep neural networks with neuron plasticity.
PATTERN RECOGNITION
(2023)
Article
Engineering, Civil
Jianchang Huang, Guohua Song, Feng He, Zhe Tan
Summary: This study examines the potential energy consumption changes in light-duty autonomous vehicles (AVs) compared with human-driving vehicles (HVs) under different traffic conditions. The research findings reveal that the energy consumption of AVs is similar to HVs at low speeds and lower at high speeds. The turning point between low-speed and high-speed varies among different data sources. Furthermore, AVs may consume more energy than HVs in complex traffic conditions at high speeds.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2023)
Article
Engineering, Electrical & Electronic
Guoqian Wu, Xin Ning, Luyang Hou, Feng He, Hengmin Zhang, Achyut Shankar
Summary: In order to enhance the classification accuracy of hyperspectral image features, it is crucial to capture the spatial spectral features. Deep learning methods have shown great promise in hyperspectral image classification due to their ability to model complex structures. In this study, a novel three-dimensional Softmax mechanism-guided bidirectional GRU network (TDS-BiGRU) is proposed for HSI classification, which outperforms several prevalent algorithms on four hyperspectral remote sensing datasets.
Proceedings Paper
Engineering, Electrical & Electronic
Tingkun Lu, Feng He
Summary: This paper proposes a radar variable scale processing method to achieve coherent integration, diversity detection, and reconstruction imaging at different scales. The method realizes scale transformation via multi-rate filter banks. Variable scale processing can resist range migration and use frequency diversity gain to find the optimal detection scale.
2022 IEEE 10TH ASIA-PACIFIC CONFERENCE ON ANTENNAS AND PROPAGATION, APCAP
(2022)