Article
Biology
Yunchou Yin, Zhimeng Han, Muwei Jian, Gai-Ge Wang, Liyan Chen, Rui Wang
Summary: In recent years, Unet and its variants have achieved remarkable success in medical image processing. However, some Unet variants increase their performance by significantly increasing the number of parameters. To address this issue, we propose a lightweight medical image segmentation network called AMSUnet, which utilizes atrous multi-scale (AMS) convolution. Our model only requires 2.62 M parameters while achieving excellent segmentation performance for small, medium, and large-scale targets.
COMPUTERS IN BIOLOGY AND MEDICINE
(2023)
Article
Acoustics
Amlu Anna Joshy, Rajeev Rajan
Summary: This paper explores the effectiveness of using the multi-head attention mechanism and multi-task learning approach for automated dysarthria severity level classification. Dysarthric speech utterances are represented by mel spectrograms and fed to a residual convolutional neural network for feature learning. The proposed method demonstrates its potential for dysarthria severity classification by achieving a gain of 11.51% and 11.58% in classification accuracy over the baseline system under speaker-dependent and speaker-independent scenarios respectively.
SPEECH COMMUNICATION
(2023)
Article
Environmental Sciences
Hao Shi, Guo Cao, Zixian Ge, Youqiang Zhang, Peng Fu
Summary: The study introduces a double-branch network that incorporates a novel pyramidal convolution and an iterative attention mechanism for extracting spectral-spatial features from hyperspectral images. Experimental results demonstrate that the model shows competitive performance in classification.
Article
Plant Sciences
Saleh Albahli, Momina Masood
Summary: Maize leaf disease has a significant impact on the quality and overall crop yield. This study introduces an end-to-end learning CNN architecture called Efficient Attention Network (EANet) based on the EfficientNetv2 model to identify various maize crop diseases. The inclusion of a spatial-channel attention mechanism allows EANet to accurately recognize multiple diseases and handle background noise in realistic field conditions. Experimental results demonstrate that the EANet model outperforms conventional CNNs in terms of performance.
FRONTIERS IN PLANT SCIENCE
(2022)
Article
Computer Science, Information Systems
Sang-woo Lee, Ryong Lee, Min-seok Seo, Jong-chan Park, Hyeon-cheol Noh, Jin-gi Ju, Rae-young Jang, Gun-woo Lee, Myung-seok Choi, Dong-geol Choi
Summary: Multi-task learning is an efficient method to tackle multiple tasks with a single model, but recent approaches struggle to outperform single-task learning. This study validates the effectiveness of MTL in low-data conditions and proposes a feature filtering module with minimal overheads. Empirical results demonstrate that MTL can significantly enhance performance under low-data conditions for all tasks.
Article
Computer Science, Information Systems
Bo Du, Ziyi Liu, Fulin Luo
Summary: RNA-binding proteins play a significant role in biological processes and mutations can lead to serious diseases. To improve prediction performance, a deep multi-scale attention network (DeepMSA) based on convolutional neural network has been proposed, outperforming several state-of-the-art methods in predicting RBPs and binding motifs.
INFORMATION SCIENCES
(2022)
Article
Fisheries
Diogo Nunes Goncalves, Plabiany Rodrigo Acosta, Ana Paula Marques Ramos, Lucas Prado Osco, Danielle Elis Garcia Furuya, Michelle Tais Garcia Furuya, Jonathan Li, Jose Marcato Junior, Hemerson Pistori, Wesley Nunes Goncalves
Summary: This study proposes a new method for locating and counting fingerlings in a sequence of images using convolutional neural networks. The method employs a multi-task approach and utilizes temporal information to enhance the results. Experimental results demonstrate that the proposed method performs well in various scenarios and is capable of detecting fingerling contact.
Article
Engineering, Electrical & Electronic
Dinghao Fan, Hengjie Lu, Shugong Xu, Shan Cao
Summary: This study introduces an end-to-end multi-task learning framework that utilizes depth modality to enhance the accuracy of gesture recognition. Experimental results demonstrate that the proposed method outperforms existing gesture recognition frameworks on three public datasets, and also achieves excellent accuracy improvement when applied to other 2D CNN-based frameworks.
IEEE SENSORS JOURNAL
(2021)
Article
Biology
Min Luo, Yi-ting Wang, Xiao-kang Wang, Wen-hui Hou, Rui-lu Huang, Ye Liu, Jian-qiang Wang
Summary: This research proposes a new deep neural network model for predicting the medical costs of diabetes. The model takes into account the multi-granularity information of medical concepts and the time interval characteristics of patients' visit sequences, and adjusts the weights of medical features using a channel attention mechanism. The superiority of the model is demonstrated through a case study and experiments.
COMPUTERS IN BIOLOGY AND MEDICINE
(2022)
Article
Computer Science, Artificial Intelligence
Jingying Chen, Lei Yang, Lei Tan, Ruyi Xu
Summary: This paper proposes a novel orthogonal channel attention-based multi-task learning approach for multi-view facial expression recognition. By utilizing a Siamese CNN and a multi-task learning framework, as well as designing a separated channel attention module and an orthogonal channel attention loss, this approach achieves good recognition accuracy on two datasets.
PATTERN RECOGNITION
(2022)
Article
Computer Science, Artificial Intelligence
Wei Chen, Ke Shi
Summary: With the rapid increase of data availability, a novel deep learning model MACNN has been developed to solve the TSC problem, achieving the best performance on 85 UCR standard datasets and outperforming other methods by a large margin.
Article
Chemistry, Multidisciplinary
Xuezhu Lin, Shihan Chao, Dongming Yan, Lili Guo, Yue Liu, Lijuan Li
Summary: This study proposes a multi-sensor data fusion method based on a self-attention mechanism to enhance data integrity and accuracy. Experimental results demonstrate that the proposed method outperforms other methods in terms of accuracy, generalization capability, and robustness, indicating its importance and application value in multi-sensor data fusion processing.
APPLIED SCIENCES-BASEL
(2023)
Article
Agriculture, Multidisciplinary
Xue Zhao, Kaiyu Li, Yunxia Li, Juncheng Ma, Lingxian Zhang
Summary: This study developed a new vegetable disease identification model, DTL-SE-ResNet50, and compared it with other models. The results showed that the new model had high identification precision and fast identification speed.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2022)
Article
Computer Science, Software Engineering
Qian Luo, Jie Shao, Wanli Dang, Long Geng, Huaiyu Zheng, Chang Liu
Summary: In this work, we propose an efficient multi-scale channel attention network (EMCA) to address the challenges of occlusion and similar appearance in person re-identification. The EMCA incorporates a novel cross-channel attention module (CCAM) that includes local cross-channel interaction (LCI) and channel weight integration (CWI). Experimental results on popular person re-identification datasets demonstrate that our EMCA outperforms existing state-of-the-art methods consistently.
Article
Multidisciplinary Sciences
Wang Gaihua, Lin Jinheng, Cheng Lei, Dai Yingying, Zhang Tianlun
Summary: Instance segmentation, a more challenging task than object detection and semantic segmentation, is crucial for scene understanding in areas like robotics, automatic driving, and medical care. This paper proposes a Hybrid Kernel Mask R-CNN, an instance segmentation method with multi-scale attention, to address issues such as low detection efficiency for low-resolution objects and slow detection speed for complex background images. The method combines different kernels and groups to extract rich information, and assigns weights to these kernels based on a multi-scale attention mechanism. Experimental results demonstrate the superiority of the proposed method over existing approaches on multiple datasets.
Article
Computer Science, Interdisciplinary Applications
Zhongyi Han, Benzheng Wei, Stephanie Leung, Ilanit Ben Nachum, David Laidley, Shuo Li
Article
Biology
Tianyang Li, Benzheng Wei, Jinyu Cong, Yanfei Hong, Shuo Li
COMPUTERS IN BIOLOGY AND MEDICINE
(2020)
Article
Computer Science, Artificial Intelligence
Yanfei Hong, Benzheng Wei, Zhongyi Han, Xiang Li, Yuanjie Zheng, Shuo Li
Article
Computer Science, Interdisciplinary Applications
Zhongyi Han, Benzheng Wei, Yanfei Hong, Tianyang Li, Jinyu Cong, Xue Zhu, Haifeng Wei, Wei Zhang
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2020)
Article
Computer Science, Artificial Intelligence
Zhongyi Han, Benzheng Wei, Xiaoming Xi, Bo Chen, Yilong Yin, Shuo Li
Summary: The paper introduces a neural-symbolic learning framework that combines deep neural learning and symbolic logical reasoning to achieve human-like learning processes for spine medical report generation.
MEDICAL IMAGE ANALYSIS
(2021)
Article
Medical Informatics
Yue Liu, Xiang Li, Tianyang Li, Bin Li, Zhensong Wang, Jie Gan, Benzheng Wei
Summary: This study proposed a novel segmentation correction algorithm to estimate lesion areas, achieving precise segmentation and recognition of small lesions such as FCI and LACI. The method showed potential clinical application and future directions include collecting more clinical data and testing more types of tiny lesions simultaneously.
BMC MEDICAL INFORMATICS AND DECISION MAKING
(2021)
Article
Computer Science, Information Systems
Yanfei Hong, Guisheng Zhang, Benzheng Wei, Jinyu Cong, Yunfeng Xu, Kuixing Zhang
Summary: This study proposes a weakly supervised semantic segmentation algorithm for precise segmentation of skin cancer lesions at different stages, achieving effective discrimination and accurate segmentation of the lesion regions through the combination of convolutional neural network and superpixel algorithm.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Biotechnology & Applied Microbiology
Guang Zhang, Xueying He, Delin Li, Cuihuan Tian, Benzheng Wei
Summary: This study demonstrates the effectiveness of tongue image in screening COVID-19 and proposes a multimodal screening method based on tongue image, chest CT, or X-ray images, which can improve the screening accuracy. This provides a new perspective and solution for controlling the pandemic.
BIOMED RESEARCH INTERNATIONAL
(2022)
Article
Computer Science, Artificial Intelligence
Hao Hou, Jun Xu, Yingkun Hou, Xiaotao Hu, Benzheng Wei, Dinggang Shen
Summary: This paper proposes a Semi-Cycled Generative Adversarial Network (SCGAN) for real-world face super-resolution (SR). By establishing two independent degradation branches and sharing the same restoration branch, SCGAN is able to effectively utilize the generative capability of GAN for accurate and robust face SR. Experimental results demonstrate that SCGAN outperforms state-of-the-art methods in recovering face structures/details and quantitative metrics for real-world face SR.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2023)
Article
Biochemistry & Molecular Biology
Yiwen Wang, Fen Yang, Dongliang Yan, Yalin Zeng, Benzheng Wei, Jianzhong Chen, Weikai He
Summary: In this study, the mechanism of BACE1 identification for three inhibitors was comparatively determined using molecular dynamics simulations and binding free energy calculations. The presence of the inhibitors affects the structural stability, flexibility, and internal dynamics of BACE1. The binding free energy calculations reveal that hydrophobic interactions play a decisive role in the binding between the inhibitors and BACE1. Residue-based free energy decomposition analysis suggests that specific residues contribute significantly to the inhibitor-BACE1 binding, providing insights for future drug design for AD treatment.
Article
Biochemistry & Molecular Biology
Fen Yang, Yiwen Wang, Dongliang Yan, Zhongtao Liu, Benzheng Wei, Jianzhong Chen, Weikai He
Summary: This study explores the binding mechanism of three inhibitors (W8Y, W8V, and W8S) to HSP90 through molecular dynamics simulations and MM-GBSA calculations. The presence of inhibitors impacts the structural flexibility, correlated movements, and dynamics behavior of HSP90. The MM-GBSA calculations reveal that van der Waals interactions are the main forces that determine inhibitor-HSP90 binding, while hydrogen-bonding interactions and hydrophobic interactions play important roles in HSP90-inhibitor identifications. Certain residues are recognized as hot spots of inhibitor-HSP90 binding, providing significant target sites for drug design related to HSP90.
Article
Computer Science, Information Systems
Tianyang Li, Benzheng Wei, Jinyu Cong, Xuzhou Li, Shuo Li
Proceedings Paper
Computer Science, Artificial Intelligence
Benzheng Wei, Zhongyi Han, Xueying He, Yilong Yin
2017 2ND IEEE INTERNATIONAL CONFERENCE ON CLOUD COMPUTING AND BIG DATA ANALYSIS (ICCCBDA 2017)
(2017)
Article
Biotechnology & Applied Microbiology
Guang Zhang, Xueying He, Delin Li, Cuihuan Tian, Benzheng Wei
Summary: Artificial intelligence-powered screening systems for COVID-19 are in urgent demand due to the global outbreak. This study demonstrates the effectiveness of using tongue image in improving the screening accuracy of COVID-19 based on chest CT or X-ray images. The results show significant improvement in classification accuracy and provide a novel solution for large-scale screening.
BIOMED RESEARCH INTERNATIONAL
(2022)