检索结果-内蒙古大学图书馆

An improved pulse coupled neural networks model for semantic IoT

Digital Communications and Networks 2024年第3期10卷 557-567页

作者： Rong Ma Zhen Zhang Yide Ma Xiping Hu Edith C.H.Ngai Victor C.M.Leung School of Information Science and Engineering Lanzhou UniversityLanzhou 730000China School of Medical Technology Beijing Institute of TechnologyBeijing100081China Department of Electrical and Electronic Engineer The University of Hong KongHong Kong 999077China College of Computer Science and Software Engineering Shenzhen UniversityShenzhen 518060China Department of Electrical and Computer Engineering the University of British ColumbiaVancouverBC V6T 1Z4Canada

In recent years,the Internet of Things(IoT)has gradually developed applications such as collecting sensory data and building intelligent services,which has led to an explosion in mobile data ***,with the rapid development of artificial intelligence,semantic communication has attracted great attention as a new communication ***,for IoT devices,however,processing image information efficiently in real time is an essential task for the rapid transmission of semantic *** the increase of model parameters in deep learning methods,the model inference time in sensor devices continues to *** contrast,the Pulse Coupled Neural Network(PCNN)has fewer parameters,making it more suitable for processing real-time scene tasks such as image segmentation,which lays the foundation for real-time,effective,and accurate image ***,the parameters of PCNN are determined by trial and error,which limits its *** overcome this limitation,an Improved Pulse Coupled Neural Networks(IPCNN)model is proposed in this *** IPCNN constructs the connection between the static properties of the input image and the dynamic properties of the neurons,and all its parameters are set adaptively,which avoids the inconvenience of manual setting in traditional methods and improves the adaptability of parameters to different types of *** segmentation results demonstrate the validity and efficiency of the proposed self-adaptive parameter setting method of IPCNN on the gray images and natural images from the Matlab and Berkeley Segmentation *** IPCNN method achieves a better segmentation result without training,providing a new solution for the real-time transmission of image semantic information.

关键词： Internet of things(IoT) Semantic information Real-time application Improved pulse coupled neural network Image segmentation

来源：评论

学校读者我要写书评

暂无评论

Revisiting LARS for Large Batch Training Generalization of Neural Networks

IEEE Transactions on Artificial Intelligence

引用

IEEE Transactions on Artificial Intelligence 2025年第5期6卷 1321-1333页

作者： Do, Khoi Nguyen, Minh-Duong Hoa, Nguyen Tien Tran-Thanh, Long Tran, Nguyen H. Pham, Quoc-Viet Trinity College Dublin School of Computer Science and Statistics Dublin 2 D02PN40 Ireland Pusan National University Department of Information Convergence Engineering Busan46241 Korea Republic of Hanoi University of Science and Technology School of Electrical and Electronic Engineering Hanoi100000 Viet Nam University of Warwick School of Computer Science CoventryCV4 7AL United Kingdom The University of Sydney School of Computer Science DarlingtonNSW2006 Australia Kyung Hee University Department of Computer Science and Engineering Yongin446-701 Korea Republic of

This article investigates large batch training techniques using layer-wise adaptive scaling ratio (LARS) across diverse settings. In particular, we first show that a state-of-the-art technique, called LARS with the warm-up, tends to be trapped in sharp minimizers early on due to redundant ratio scaling. Additionally, a fixed steep decline in the latter phase restricts deep neural networks from effectively navigating early-phase sharp minimizers. To address these issues, we propose time varying LARS (TVLARS), a novel algorithm that replaces warm-up with a configurable sigmoid-like function for robust training in the initial phase. TVLARS promotes gradient exploration early on, surpassing sharp optimizers and gradually transitioning to LARS for robustness in later stages. Extensive experiments demonstrate that TVLARS consistently outperforms LARS and LAMB in most cases, with up to 2% improvement in classification scenarios. In all self-supervised learning cases, TVLARS achieves up to 10% performance improvement. © 2020 IEEE.

关键词： Self-supervised learning

来源：评论

学校读者我要写书评

暂无评论

Evolutionary Optimization with a Simplified Helper Task for High-Dimensional Expensive Multiobjective Problems

引用

ACM Transactions on Evolutionary Learning and Optimization 2024年第4期4卷 1-32页

作者： Wu, Xunfeng Lin, Qiuzhen Zhou, Junwei Liu, Songbai Coello Coello, Carlos A. Leung, Victor C. M. College of Computer Science and Software Engineering Shenzhen University Shenzhen518060 China School of Computer Science and Artificial Intelligence Wuhan University of Technology Wuhan430070 China Mexico City07300 Mexico Department of Electrical and Computer Engineering University of British Columbia VancouverBCV6T 1Z4 Canada

In recent years, surrogate-Assisted evolutionary algorithms (SAEAs) have been sufficiently studied for tackling computationally expensive multiobjective optimization problems (EMOPs), as they can quickly estimate the qualities of solutions by using surrogate models to substitute for expensive evaluations. However, most existing SAEAs only show promising performance for solving EMOPs with no more than 10 dimensions, and become less efficient for tackling EMOPs with higher dimensionality. Thus, this article proposes a new SAEA with a simplified helper task for tackling high-dimensional EMOPs. In each generation, one simplified task will be generated artificially by using random dimension reduction on the target task (i.e., the target EMOPs). Then, two surrogate models are trained for the helper task and the target task, respectively. Based on the trained surrogate models, evolutionary multitasking optimization is run to solve these two tasks so that the experiences of solving the helper task can be transferred to speed up the convergence of tackling the target task. Moreover, an effective model management strategy is designed to select new promising samples for training the surrogate models. When compared to five competitive SAEAs on four well-known benchmark suites, the experiments validate the advantages of the proposed algorithm on most test cases. © 2024 held by the owner/author(s).

关键词： Multitasking

来源：评论

学校读者我要写书评

暂无评论

Virtual Reduced-Order Model-Based Back EMF Estimation and Speed Sensorless Control for LC-Filtered PMSM Drives

引用

IEEE Transactions on Power Electronics 2025年第7期40卷 9749-9763页

作者： Xue, Cheng Wu, Xuesong Li, Yuzhuo Li, Yunwei Southwest Jiaotong University School of Integrated Circuits Science and Engineering Chengdu611756 China University of Alberta Department of Electrical and Computer Engineering EdmontonABT6G 2V4 Canada

The installment of the LC filter at the inverter output side reshapes the sinusoidal input voltage for the motor terminal, thus, extending a longer motor lifetime. Despite this, achieving speed sensorless control remains essential for enhanced reliability and saved costs. Currently, there is limited research on the development of a speed sensorless control tailored to high-order drive scenarios due to the increased complexity and strong coupling of the system modeling, and the unaltered adoption of the prevailing general observer methodology demands a considerably large dimensional gain matrix to guarantee observability. To fill this important research gap, this article proposes a novel back-electromotive force (EMF) modeling for the permanent magnet synchronous machine drives based on the weighted current between the filter inductor current and motor stator current. This new approach converts the third-order LCL model to the first-order L case (virtual reduced-order modeling), which enables the estimation of back EMF without relying on voltage sensors. In addition, the observer designed based on the proposed virtual model not only removes the dependency on capacitor parameters but also reduces the size of the gain matrix. This further enhances the robustness of the back EMF estimation and simultaneously simplifies the design and computation of the observation algorithm. A Kalman filter observer for back EMF estimation is implemented as a case study to verify the proposed modeling. The efficacy of the proposed speed sensorless control is also evaluated under scenarios involving variations in filter and motor inductance. © 1986-2012 IEEE.

关键词： Permanent magnets

来源：评论

学校读者我要写书评

暂无评论

Truncation and Rounding-Based Scalable Approximate Multiplier Design for computer Imaging Applications

引用

computers, Materials & Continua 2022年第12期73卷 5169-5184页

作者： S.Rooban A.Yamini Naga Ratnam M.V.S.Ramprasad N.Subbulakshmi R.Uma Mageswari Department of Electronics and Communication Engineering Koneru Lakshmaiah Education FoundationVaddeswaramGuntur522502Andhra PradeshIndia Department of EECE GITAM(Deemed to be University)VisakhapatnamAPIndia Department of Electronics and Communication Engineering Francis Xavier Engineering CollegeTirunelveli627003Tamil NaduIndia 4Department of Computer Science and Engineering Vardhaman College of EngineeringShamshabadHyderabad501218TelanganaIndia

Advanced technology used for arithmetic computing application,comprises greater number of approximatemultipliers and approximate *** and Rounding-based Scalable ApproximateMultiplier(TRSAM)distinguish a variety of modes based on height(h)and truncation(t)as TRSAM(h,t)in the *** TRSAM operation produces higher absolute error in Least Significant Bit(LSB)data shift unit.A new scalable approximate multiplier approach that uses truncation and rounding TRSAM(3,7)is proposed to increase themultiplier *** the help of foremost one bit architecture,the proposed scalable approximate multiplier approach reduces the partial *** proposed approximate TRSAM multiplier architecture gives better results in terms of area,delay,and *** accuracy of 95.2%and the energy utilization of 24.6 nJ is observed in the proposed multiplier *** proposed approach shows 0.11%,0.23%,and 0.24%less Mean Absolute Relative Error(MARE)when compared with the existing approach for the input of 8-bit,16-bit,and 32-bit *** also shows 0.13%,0.19%,and 0.2%less Variance of Absolute Relative Error(VARE)when compared with the existing approach for the input of 8-bit,16-bit,and 32-bit *** proposed approach is implemented with Field-Programmable Gate Array(FPGA)and shows the delay of 3.640,6.481,12.505,22.572,and 36.893 ns for the input of 8-bit,16-bit,32-bit,64-bit,and 128-bit *** proposed approach is applied in digital filters designwhich shows the Peak-Signal-to-NoiseRatio(PSNR)of 25.05 dB and Structural Similarity Index Measure(SSIM)of 0.98 with 393 pJ energy consumptions when used in image *** proposed approach is simulated with Xilinx and MATLAB and implemented with FPGA.

关键词： Truncation rounding based scalable approximate multiplier foremost one detector field programmable gate array peak-signal-to-noise-ratio structural similarity index measure

来源：评论

学校读者我要写书评

暂无评论

A Comparative Analysis of the Junction Formation Mechanisms of the Boron on Silicon Junction 33

A Comparative Analysis of the Junction Formation Mechanisms ...

引用

33rd International Scientific Conference Electronics, ET 2024

作者： Fang, Piet Nihtianov, Stoyan TU Delft Faculty of Electrical Engineering Mathematics and Computer Science Department of Microelectronics Mekelweg 4 Delft Netherlands

ISBN: (纸本)9798350376449

Photodiodes based on the Boron on Silicon junction (B-Si) show excellent responsivity to DUV and VUV photons, radiation hardness, and impressive electrical characteristics. However, the proposed models describing the junction formation mechanism do not sufficiently predict the junction's properties. We analyze two previously proposed models: the ultra-shallow p-n junction model and the charge transfer heterojunction model. We additionally apply the Schottky-Mott theory, a semiconductor-metal heterojunction model. Both the commonalities and incompatibilities between these models are discussed. © 2024 IEEE.

关键词： Heterojunctions

来源：评论

学校读者我要写书评

暂无评论

CPDZ: A Credibility-Aware and Privacy-Preserving Data Collection Scheme with Zero-Trust in Next-Generation Crowdsensing Networks

引用

IEEE Journal on Selected Areas in Communications 2025年第6期43卷 2183-2199页

作者： Tang, Jianheng Fan, Kejia Yang, Shihao Liu, Anfeng Xiong, Neal N. Song, Houbing Herbert Leung, Victor C. M. Central South University School of Computer Science and Engineering Changsha410083 China Peking University School of Computer Science Beijing100871 China Sul Ross State University Department of Computer Mathematical and Physical Sciences AlpineTX79830 United States Department of Information Systems BaltimoreMD21250 United States Shenzhen MSU-BIT University Artificial Intelligence Research Institute Shenzhen518172 China Shenzhen University College of Computer Science and Software Engineering Shenzhen518060 China The University of British Columbia Department of Electrical and Computer Engineering VancouverBCV6T 1Z4 Canada

Next-Generation Crowdsensing Networks (NGCNs) have become increasingly critical for smart cities, where data privacy and quality are pivotal concerns. Traditional trust mechanisms in crowdsensing mainly rely on static trust models, which are insufficient for dynamic security requirements. Zero-Trust security represents a promising opportunity, yet coming with notable challenges in NGCNs, including Unknown Workers Online Recruitment (UWOR), Information Elicitation Without Verification (IEWV), Privacy Preserving Data Evaluation (PPDE), and Dynamic Trust Abrupt Shift (DTAS). To address these challenges, we propose a Credibility-aware and Privacy-preserving Data collection scheme with Zero-trust (CPDZ) for secure and quality data collection in NGCNs. First, our CPDZ scheme encompasses a quality worker recruitment strategy with combinatorial multi-armed bandit models, utilizing Thompson Sampling for the secure and efficient resolution of the UWOR. Second, an active dispatching scheme for unmanned aerial vehicles is crafted to collect data as a gold standard to assist in overcoming the IEWV challenge. Third, as for the PPDE challenge, we propose a lightweight privacy-preserving scheme for dependable truth discovery and secure trust verification. Fourth, the DTAS challenge is managed by a dual verification scheme that integrates short-term and long-term trust assessments, ensuring stability and adaptability of the zero-trust security in our CPDZ scheme. Experiments confirm the superiority of our CPDZ scheme, showing a 12.5% increase in recruitment revenue and a 57.8% reduction in relative error compared to existing approaches. © 1983-2012 IEEE.

关键词： Differential privacy

来源：评论

学校读者我要写书评

暂无评论

Enhancing breast cancer histopathological image classification using attention-based high order covariance pooling

引用

Neural Computing and Applications 2024年第36期36卷 23275-23293页

作者： Waqas, Muhammad Ahmed, Amr Maul, Tomas Liao, Iman Yi School of Computer Science Faculty of Science and Engineering University of Nottingham Malaysia Campus Selangor Dar ul Ehsan Semenyih43500 Malaysia Department of Software Engineering University of Malakand Khyber Pakhtunkhwa Lower Dir Chakdara18800 Pakistan Department of Computer Science Edge Hill University Lancashire L39 4QP United Kingdom School of Computer Science Faculty of Science University of Nottingham NottinghamNG8 1BB United Kingdom

Breast cancer, the most common cancer affecting female patients, presents serious challenges for proper detection. Although computer-aided diagnostic techniques have progressed, their accuracy and efficacy remain limited. To overcome these challenges, we introduce DHA-Net, a new deep learning system that combines an effective attention module (EAM) and a high-order pooling layer with a ResNet-18 backbone. DHA-Net is tested using three well-known breast cancer histopathology image datasets: BreakHis, BACH2018, and a closely related Kaggle-Breast cancer histopathology dataset. Our experiments show that DHA-Net not only improves on existing state-of-the-art approaches, but significantly outperforms them in classifying breast cancer images. This work emphasizes the novel combination of an EAM with high-order pooling, demonstrating DHA-Net’s potential to improve diagnostic accuracy and serve as a more effective tool for medical imaging applications. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2024.

关键词： Lung cancer

来源：评论

学校读者我要写书评

暂无评论

Towards Robust Cyber Attack Taxonomies: A Survey with Requirements, Structures, and Assessment

引用

ACM Computing Surveys 2025年第8期57卷 1-36页

作者： Santos, Paulo Roberto Da Paz Ferraz Resende, Paulo Angelo Alves Gondim, João José Costa Drummond, André Costa Department of Computer Science University of Brasilia Brasilia Brazil C4I and Cyber Center George Mason University Fairfax United States GigaCandanga Association Brasilia Brazil Department of Electrical Engineering University of Brasilia Brasilia Brazil Institute of Computer and Network Engineering Technische Universität Braunschweig Braunschweig Germany

Cyber attacks have become a growing threat in today's interconnected society, and taxonomies play a crucial role in understanding and preventing these attacks. However, the lack of comprehensive assessment methods for evaluating attack taxonomies represents a significant gap in the literature, hindering their development and applicability. This article aims to address this gap by conducting a survey of 20 attack taxonomies published between 2011 and 2022 and evaluating them with a novel set of qualitative and quantitative assessment criteria, grounded in fundamental taxonomy requirements and key structural attributes. In pursuit of clear and objective assessment criteria, the authors investigated the main taxonomy properties in the literature, identifying dependencies and relationships. This investigation extracted the fundamental requirements for a relevant and widely accepted attack taxonomy in the cybersecurity community. Noteworthy structural aspects, such as organization, scheme, labeling, and approach, are also addressed, considering their impact on taxonomy effectiveness and applicability constraints. Finally, the article poses some open questions and challenges, along with suggestions for future research directions. © 2025 Copyright held by the owner/author(s). Publication rights licensed to ACM.

关键词： Cyber attacks

来源：评论

学校读者我要写书评

暂无评论

Intelligent Proximate Analysis of Coal Based on Near-Infrared Spectroscopy and Multioutput Deep Learning

IEEE Transactions on Artificial Intelligence

引用

IEEE Transactions on Artificial Intelligence 2024年第3期5卷 1398-1410页

作者： Zou, Liang Qiao, Jiahui Yu, Xinhui Chen, Xun Lei, Meng School of Information and Control Engineering China University of Mining and Technology Xuzhou221116 China Department of Electrical and Computer Engineering University of British Columbia VancouverBCV6T 1Z4 Canada Department of Electronic Engineering and Information Science University of Science and Technology of China Hefei230027 China

Proximate analysis of coal indicates the moisture, ash, volatile content, and calorific value, which has been widely utilized as the basis for coal characterization. It involves heating the coal under various conditions until a constant weight is obtained. Although it is a relatively simple process that does not require expensive analytical equipment, determining these characteristics is time consuming. An alternative way for proximate analysis is spectral analysis in combination with various machine learning methods. However, most previous works analyze individual characteristics and fail to explore the relationship among them. In this study, we propose a method for proximate analysis based on near-infrared spectroscopy and a multioutput attention Unet (MOA-Unet), which can predict multiple characteristics simultaneously. First, an attention-based Unet is designed as the shared feature extraction subnetwork, including an encoder, a decoder, convolutional block attention modules, and multiscale feature fusion modules, which can improve the representation power of the U-shape network through aggregating features of shallower layers and concatenating features of deeper layers. Second, four individual subnetworks with fully connected layers, designed for four outputs, are utilized for regressing those four characteristics. We employ the gradient normalization algorithm to alleviate the gradient magnitude masking effect caused by training imbalance among different tasks. The proposedMOA-Unet is compared with classical chemometric methods on 670 coal samples from on-site *** experimental results demonstrate that the proposedmodel achieves state-of-the-art performance with correlation coefficients of 0.9015, 0.9538, 0.8986, and 0.8884, corresponding to moisture, ash, volatile content, and calorific value, respectively. Impact Statement-The proximate analysis of coal has been widely utilized as the basis for determining the rank of coal which is in connection with coa

关键词： Feature extraction

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：