检索结果-内蒙古大学图书馆

Spike-Based Neuromorphic Model of Spasticity for Generation of Affected Neural Activity

IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING 2025年 33卷 1360-1371页

作者： Yan, Jin Luo, Qi Chen, Zhi Wang, Zeyu Guo, Xinliang Xie, Qing Oetomo, Denny Tan, Ying Niu, Chuanxin M. Shanghai Jiao Tong Univ Ruijin Hosp Sch Med Dept Rehabil Med Shanghai 200025 Peoples R China Changsha Univ Sci & Technol Sch Automot & Mech Engn Changsha 410000 Peoples R China Shanghai Ruijin Rehabil Hosp Dept Rehabil Med Shanghai 200023 Peoples R China Univ Melbourne Dept Mech Engn Parkville Vic 3010 Australia

is a common motor symptom that disrupt muscle contraction and hence movements. Proper management of spasticity requires identification of its origins and reasoning of the therapeutic plans. Challenges arise because spasticity might originate from elevated activity in both the cortical and sub-cortical pathways. No existing models (animal or computational) could cover all possibilities leading to spasticity, especially the peripheral causes such as hyperreflexia. To bridge this gap, this work develops a novel computational, spike-based neuromorphic model of spasticity, named NEUSPA. Rather than relying solely on a monosynaptic spinal loop comprising alpha motoneurons, sensory afferents, synapses, skeletal muscles, and muscle spindles, the NEUSPA model introduces two additional inputs: additive (ADD) and multiplicative (MUL). These inputs generate velocity-dependent EMG responses. The effectiveness of the NEUSPA model is validated using classic experiments from the literature and data collected from two post-stroke patients with affected upper-limb movements. The model is also applied to simulate two real-world scenarios that patients may encounter. Simulation results suggest that hyperreflexia due to extra inputs was sufficient to produce spastic EMG responses. However, EMG onsets were more sensitive to ADD inputs (slope = 0.628, p < 0.0001, r(2) = 0.96)compared to MUL inputs (slope = 0.471, p < 0.0001, r2= 0.92). Addition ally, simulation of finger-pressing on a deformable object indicated that spasticity could increase the duration from 1.03s to 1.20s compared to a non-impaired condition. These results demonstrate that NEUSPA effectively synthesizes abnormal physiological data, facilitating decision-making and machine learning in neurorehabilitation.

关键词： Muscles Spasticity computational modeling Electromyography Integrated circuit modeling Animals Synapses Neurons Neuromorphics Motors neuromorphic modeling spiking neurons electromyography

来源：评论

学校读者我要写书评

暂无评论

Supervised Detail-Guided Multiscale State-Space Model for Pan-Sharpening

引用

IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING 2025年 63卷

作者： Zhao, Mengjiao Ma, Mengting Li, Xiangdong Ma, Xiaowen Zhang, Wei Song, Siyang Zhejiang Univ Sch Software Technol Hangzhou 310027 Peoples R China Zhejiang Univ Sch Comp Sci & Technol Hangzhou 310027 Peoples R China Zhejiang Univ Innovat Ctr Yangtze River Delta Jiaxing 314103 Zhejiang Peoples R China Univ Exeter Dept Comp Sci Exeter EX4 4PY England

Pan-sharpening reconstructs the high-resolution multispectral (HR-MS) image from its corresponding panchromatic (PAN) image and low-resolution multispectral (LR-MS) image. However, existing deep learning (DL)-based pan-sharpening methods typically suffer from three challenges: 1) the vanilla LR-MS image upsampling employed by them fails to consider domain knowledge, thereby disregarding crucial information;2) while remote sensing images exhibit multiscale complex land features, existing methods fail to fully exploit crucial multiscale spatial information, that is, scale transformation layers in their models are not effective;and 3) existing convolutional neural network (CNN) and transformer-based pan-sharpening backbones are constrained by inherent local receptive fields or quadratic computational complexity, making them difficult to balance their effectiveness and efficiency. To address these issues, we propose a novel supervised detail-guided multiscale state-space model for pan-sharpening, namely SDMSPan. Our SDMSPan consists of three residual state-space modules (Res-SSMs) that are responsible for handling image information at three spatial scales, where each Res-SSM aims to model both local and long-range dependencies between PAN and LR-MS images at a specific spatial scale with lower computational cost. Between each pair of Res-SSMs, a novel detail-guided upsampling block (DGUB) is proposed to apply spatial details of the PAN image to guide effective and task-aware intermediate feature upsampling, where a novel multiscale intermediate spatial-spectral supervision strategy is also proposed to supervise the training of every DGUB. Experimental results demonstrate that our proposed approach significantly outperforms other state-of-the-art methods in performance. Our code is provided at https://***/zhaomengjiao123/SDMSPan.

关键词： Feature extraction computational modeling Distortion Transformers Spatial resolution computational efficiency Optimization Data mining Remote sensing Visualization Pan-sharpening remote sensing state-space model

来源：评论

学校读者我要写书评

暂无评论

Multi-Functional Homomorphic Encryption Method Based on Crowd Sensing Networks

引用

IEEE ACCESS 2025年 13卷 36795-36803页

作者： Chen, Hui Mou, Xingyu Wang, Zixuan Wu, Tianting Wang, Xin Wang, Chengcheng Song, Ruixiang Song, Lifang Jiang, Xiaoling Zhang, Xinjing Li, Yunfeng Jilin Prov Meteorol Informat Network Ctr Changchun 130062 Jilin Peoples R China Changchun Meteorol Informat Technol Secur Ctr Changchun 130012 Jilin Peoples R China Jilin Prov Forestry Informat Ctr Changchun 130022 Jilin Peoples R China Jilin Prov Meteorol Serv Ctr Changchun 130062 Jilin Peoples R China Jilin Prov Meteorol Disaster Prevent Technol Ctr Changchun 130062 Jilin Peoples R China

In response to the security and privacy issues associated with sensing devices in contemporary crowd sensing systems, the paper proposes a crowd sensing networks method based on the MFHE (the Multi-Functional Homomorphic Encryption) architecture. This method leverages fully homomorphic encryption technology, thereby eliminating the use of plaintext throughout the sensing process and enhancing security compared to traditional encryption methods. Initially, a sensing model is constructed based on the classic crowd sensing architecture. Subsequently, fully homomorphic encryption technology is employed for plaintext-free data transmission. The encryption scheme is developed on the MFHE architecture, with additional improvements introduced through indistinguishability obfuscation and puncturable pseudorandom function methods, which serve to shorten the ciphertext and reduce complexity, rendering it more suitable for crowd sensing environments. Finally, experiments conducted on real-world datasets demonstrate that the proposed method is effective in terms of both privacy protection and time efficiency.

关键词： Sensors Data privacy Homomorphic encryption Protection Privacy Public key Differential privacy Data models computational modeling computational efficiency Crowd sensing networks multi-functional homomorphic encryption indistinguishability obfuscation puncturable pseudorandom

来源：评论

学校读者我要写书评

暂无评论

Boosting Cyberattack Detection Using Binary Metaheuristics With Deep Learning on Cyber-Physical System Environment

引用

IEEE ACCESS 2025年 13卷 11280-11294页

作者： Al Mazroa, Alanoud Albogamy, Fahad R. Khairi Ishak, Mohamad Mostafa, Samih M. Princess Nourah bint Abdulrahman Univ PNU Coll Comp & Informat Sci Dept Informat Syst POB 84428 Riyadh 11671 Saudi Arabia Taif Univ Turabah Univ Coll Comp Sci Program Taif 21944 Saudi Arabia Ajman Univ Coll Engn & Informat Technol Dept Elect & Comp Engn Ajman U Arab Emirates South Valley Univ Fac Comp & Informat Comp Sci Dept Qena 83523 Egypt New Assiut Technol Univ NATU Fac Ind & Energy Technol New Assiut City 71684 Egypt

The swift advancement of cyber-physical systems (CPSs) across sectors such as healthcare, transportation, critical infrastructure, and energy enhances the crucial requirement for robust cybersecurity measures to protect these systems from cyberattacks. The cyber-physical method is a hybrid of cyber and physical components, and a safety breach in the element is central to catastrophic consequences. Cyberattack recognition and mitigation techniques in CPSs include using numerous models like intrusion detection systems (IDSs), access control mechanisms, encryption, and firewalls. Cyberattack detection employing deep learning (DL) contains training neural networks to identify patterns indicative of malicious actions within system logs or network traffic, allowing positive classification and mitigation of cyber-attacks. By leveraging the integral ability of DL methods to learn complex representations, this technique enhances the accuracy and efficiency of detecting diverse and growing cyber-attacks. Thus, the study proposes an automated Cyberattack Detection using Binary Metaheuristics with Deep Learning (ACAD-BMDL) method in a CPS environment. The ACAD-BMDL method mainly focuses on enhancing security in the CPS environment via the cyberattack detection process. The ACAD-BMDL method uses Z-score normalization to scale the input dataset. In addition, the binary grey wolf optimizer (BGWO) model is utilized to choose an optimal feature subset. Moreover, the Enhanced Elman Spike Neural Network (EESNN) model detects cyber-attacks. Furthermore, the Archimedes Optimization Algorithm (AOA) model is employed to select the optimum hyperparameter for the EESNN model. The empirical analysis of the ACAD-BMDL technique is performed on a benchmark dataset. The experimental validation of the ACAD-BMDL technique portrayed a superior accuracy value of 99.12% and 99.36% under NSLKDD2015 and CICIDS2017 datasets in the CPS environment.

关键词： Cyberattack Optimization Feature extraction Data models Accuracy Security Neural networks Adaptation models Deep learning computational modeling Cyber-physical systems deep learning cyberattack detection metaheuristics Archimedes optimization algorithm

来源：评论

学校读者我要写书评

暂无评论

Human Digital Twins: Enhancing Interactions With Digital Ecosystems

引用

IEEE INTERNET COMPUTING 2025年第1期29卷 56-64页

作者： Laso, Sergio Herrera, Juan Luis Galan-Jimenez, Jaime Berrocal, Javier Global Proc & Prod Improvement SL Caceres 10004 Spain Univ Extremadura Comp Sci & Commun Engn Dept Caceres 10004 Spain Univ Extremadura Caceres 10004 Spain

The emergence of digital twins (DTs) has transformed domains like Industry 4.0 or automotive, enabling advanced insights and predictive maintenance as well as driving efficiency and innovation. With the evolution toward human-centric domains, such as Industry 5.0 or intelligent vehicles, the need for human DTs (HDTs)-human digital representations with the aim to align human interactions with the system's design and its performance-arises. HDTs, linked to human interactions in various contexts, seek to improve well-being, safety, and productivity by integrating human factors into digital systems. However, they face unique considerations and challenges compared to traditional DTs. We provide an overview of distributed multicontext HDTs, highlighting their utility in various domains, and propose an architecture designed to address key challenges in their development and deployment, including data management, privacy, or interoperability. Additionally, we show the development of HDTs in practical and functional environments through a set of artifacts.

关键词： Digital twins Artificial intelligence Productivity Fifth Industrial Revolution computational modeling Smart cities Internet Intelligent vehicles Software development management Privacy Maintenance management Predictive maintenance

来源：评论

学校读者我要写书评

暂无评论

GSSR-Net: Geo-Spatial Structural Refinement Network for Remote Sensing Change Detection

引用

IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING 2025年 63卷

作者： Wang, Shuo Yuan, Genji Li, Jinjiang Shandong Technol & Business Univ Sch Comp Sci & Technol Yantai 264005 Peoples R China

Remote sensing (RS) change detection is a technique for identifying changes in the ground surface by comparing RS images from different periods. Although such tasks have been developed for a long time and some methods have been proposed to enhance the features of real changes in the ground objects, they still face the perception ambiguity caused by the heterogeneity of the ground objects in complex RS environments: 1) insufficient processing of nonstationary changes between dual-temporal image features and 2) high spatial heterogeneity leads to difficulties in structural identification. In order to solve the interference of these problems on the downstream tasks of change detection, this article proposes geo-spatial structural refinement network (GSSR-Net) for RS change detection. First, we introduce a DualTime Mamba structure with an omnidirectional scanning path, adjust its input matrix between dual-temporal image features in the deep scale, and allow the model to fully consider the spatiotemporal dependence and image structure information of the previous and next time points. In addition, this article designs a land-cover feature extraction (LCFE) method to improve the perception ability of the ground object target structure. Specifically, this method refines the image edge by separating high frequencies, adjusts the contour structure information of the dual-phase image by separating low frequencies, and then further models the relationship between pixels by combining feature structures and spatial offset mechanisms. Our experimental results on three datasets demonstrate the superiority of GSSR-Net. The network code address is https://***/SparrowTought/GSSR-Net.

关键词： Remote sensing Feature extraction Transformers computational modeling Convolutional neural networks Convolution Accuracy Support vector machines Image edge detection Image coding Building change detection (BCD) graph convolution high-resolution remote sensing (RS) images Mamba

来源：评论

学校读者我要写书评

暂无评论

Smart Farming Solutions: A User-Friendly GUI for Maize Tassel Estimation Using YOLO With Dynamic and Fixed Labelling, Featuring Video Support

引用

IEEE ACCESS 2025年 13卷 57809-57833页

作者： Jahangir Moshayedi, Ata Wang, Zhonghua Sharifdoust, Maryam Sioofy Khoojine, Arash Zhang, Wei Kolahdooz, Amin Hu, Jiandong Jiangxi Univ Sci & Technol Sch Informat Engn Ganzhou 341000 Jiangxi Peoples R China McMaster Univ Dept Math & Stat Hamilton ON L8S 4L8 Canada Islamic Azad Univ Dept Math & Stat Khomeinishahr Branch Khomeinishahr Iran Yibin Univ Fac Econ & Business Adm Yibin 644001 Peoples R China Henan Agr Univ Dept Elect Engn Zhengzhou 450002 Henan Peoples R China De Montfort Univ Sch Engn & Sustainable Dev Fac Technol Leicester LE1 9BH Leics England

The integration of Autonomous Aerial Vehicles (AAVs) has significantly advanced image processing and remote sensing, particularly in precision agriculture. These technologies enhance data collection and agricultural yield estimation, benefiting banks, insurance companies, and government agencies in decision-making for budget allocation and quality assessments. This study addresses the challenge of accurately quantifying corn production by developing an enhanced YOLO-v8-based deep learning model, incorporating dynamic and fixed labeling techniques, tested on 810 images and video data for real-time detection. The research utilized two primary datasets totaling 570 images. The evaluation process comprised four distinct tests: Test 1, conducted on Dataset 1 with 200 images, assessed seven attention mechanisms (SE, CBAM, GA, LKA, CA, SA, and TA) using deep learning metrics (Precision, Recall, mAP50, mAP50-95, F1-score) and statistical methods (Duncan's test). Test 2 validated model performance on 370 images from external sources, where *** achieved 97.48% accuracy, outperforming *** (95.13%). Test 3, comparing with the MTDC benchmark dataset, confirmed ***'s accuracy at 95.93%, exceeding previous reports, while *** achieved 95.71%. Finally, Test 4, utilizing video-based evaluation via a developed GUI, demonstrated ***'s superiority (95.77%) over *** (95.48%) and YOLO-v5 (95.72%), significantly outperforming the standard YOLO model (72.79%). This study advances computer vision in agriculture, offering a scalable, high-accuracy model for corn yield estimation, with broad applications in farming optimization, financial planning, and policy-making.

关键词： Accuracy Production Yield estimation Kernel Convolutional neural networks YOLO Remote sensing Deep learning Data models computational modeling Precise agriculture corn estimation autonomous aerial vehicle (AAV) YOLO-v8 YOLO.LKA YOLO.SA remote sensing

来源：评论

学校读者我要写书评

暂无评论

Blind Video Quality Assessment at the Edge

引用

IEEE TRANSACTIONS ON MULTIMEDIA 2025年 27卷 2320-2334页

作者： Mei, Zhanxuan Wang, Yun-Cheng Kuo, C. -C. Jay Univ Southern Calif Ming Hsieh Dept Elect & Comp Engn Los Angeles CA 90089 USA

Owing to the proliferation of user-generated videos on the Internet, blind video quality assessment (BVQA) at the edge attracts growing attention. The usage of deep-learning-based methods is restricted to be applied at the edge due to their large model sizes and high computational complexity. In light of this, a novel lightweight BVQA method called GreenBVQA is proposed in this work. GreenBVQA features a small model size, low computational complexity, and high performance. Its processing pipeline includes: video data cropping, unsupervised representation generation, supervised feature selection, and mean-opinion-score (MOS) regression and ensembles. We conduct experimental evaluations on three BVQA datasets and show that GreenBVQA can offer state-of-the-art performance in the Pearson Linear Correlation Coefficient (PLCC) and the Spearman Rank Order Correlation Coefficient (SROCC) metrics while demanding significantly smaller model sizes and lower computational complexity. Thus, GreenBVQA is well-suited for edge devices.

关键词： Streaming media Quality assessment Feature extraction Video recording computational modeling Edge computing Prediction algorithms Green products computational complexity Quality of experience Video quality assessment blind video quality assessment edge computing green learning

来源：评论

学校读者我要写书评

暂无评论

GBCT: Efficient and Adaptive Clustering via Granular-Ball Computing for Complex Data

引用

IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS 2025年第6期36卷 9961-9969页

作者： Xia, Shuyin Shi, Bolun Wang, Yifan Xie, Jiang Wang, Guoyin Gao, Xinbo Chongqing Univ Posts & Telecommun Chongqing Key Lab Computat Intelligence Key Lab Cyberspace Big Data Intelligent Secur Minist Educ Chongqing 400065 Peoples R China Chongqing Univ Posts & Telecommun Key Lab Big Data Intelligent Comp Chongqing 400065 Peoples R China Chongqing Univ Posts & Telecommun Dept Comp Sci Chongqing 400065 Peoples R China

Traditional clustering algorithms often focus on the most fine-grained information and achieve clustering by calculating the distance between each pair of data points or implementing other calculations based on points. This way is not inconsistent with the cognitive mechanism of "global precedence" in the human brain, resulting in those methodsbad performance in efficiency, generalization ability, and robustness. To address this problem, we propose a new clustering algorithm called granular-ball clustering via granular-ball computing. First, clustering algorithm based on granular-ball (GBCT) generates a smaller number of granular-balls to represent the original data and forms clusters according to the relationship between granular-balls, instead of the traditional point relationship. At the same time, its coarse-grained characteristics are not susceptible to noise, and the algorithm is efficient and robust;besides, as granular-balls can fit various complex data, GBCT performs much better in nonspherical datasets than other traditional clustering methods. The completely new coarse granularity representation method of GBCT and cluster formation mode can also be used to improve other traditional methods.

关键词： Clustering algorithms Partitioning algorithms computational modeling Robustness Noise Approximation algorithms Mathematical models Clustering methods Telecommunications Shape Clustering granular computing granular-ball multigranularity

来源：评论

学校读者我要写书评

暂无评论

The Convex Landscape of Neural Networks: Characterizing Global Optima and Stationary Points via Lasso Models

引用

IEEE TRANSACTIONS ON INFORMATION THEORY 2025年第5期71卷 3854-3870页

作者： Ergen, Tolga Pilanci, Mert LG AI Res Ann Arbor MI 48105 USA Stanford Univ Elect Engn Dept Stanford CA 94305 USA

Due to the non-convex nature of training Deep Neural Network (DNN) models, their effectiveness relies on the use of non-convex optimization heuristics. Traditional methods for training DNNs often require costly empirical methods to produce successful models and do not have a clear theoretical foundation. In this study, we examine the use of convex optimization theory and sparse recovery models to refine the training process of neural networks and provide a better interpretation of their optimal weights. We focus on training two-layer neural networks with piecewise linear activations and demonstrate that they can be formulated as a finite-dimensional convex program. These programs include a regularization term that promotes sparsity, which constitutes a variant of group Lasso. We first utilize semi-infinite programming theory to prove strong duality for finite width neural networks and then we express these architectures equivalently as high dimensional convex sparse recovery models. Remarkably, the worst-case complexity to solve the convex program is polynomial in the number of samples and number of neurons when the rank of the data matrix is bounded, which is the case in convolutional networks. To extend our method to training data of arbitrary rank, we develop a novel polynomial-time approximation scheme based on zonotope subsampling that comes with a guaranteed approximation ratio. We also show that all the stationary points of the nonconvex training objective can be characterized as the global optimum of a subsampled convex program. Our convex models can be trained using standard convex solvers without resorting to heuristics or extensive hyper-parameter tuning unlike non-convex methods. Due to the convexity, optimizer hyperparameters such as initialization, batch sizes, and step size schedules have no effect on the final model. Through extensive numerical experiments, we show that convex models can outperform traditional non-convex methods and are not sensitive

关键词： Training Biological neural networks Convex functions Vectors Standards Neurons Data models Minimization Data mining computational modeling Neural networks convex optimization polynomial-time complexity piecewise linear activations stationary points implicit regularization

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：