检索结果-内蒙古大学图书馆

Yolov5s-PSG: Improved Yolov5s-Based Helmet Recognition in Complex Scenes

IEEE ACCESS 2025年 13卷 34915-34924页

作者： Li, Yi Shi, Xiaolin Xu, Xiaolong Zhang, Han Yang, Fan Liaoning Univ Technol Coll Mech Engn & Automat Jinzhou 121001 Peoples R China Northwestern Polytech Univ Sch Mech Engn Inst Intelligent Mfg Xian 710072 Peoples R China

In the field of industrial safety, due to the existence of color, distance and other reasons in complex industrial environments caused by the helmet small target detection methods have the problem of misdetection and omission, and the Yolov5s model for real-time detection of helmets is not ideal. for the purpose of Improvement of extraction capabilities multi-scale features, improve the network's focus on key features, and achieve an excellent balanced effect for accuracy and speed, the model Yolov5s-PSG based on Yolov5s improvement is proposed. First, the C3_PPA module improves the detection accuracy through a multi-branch feature extraction strategy and an attention mechanism. Second, the SimAM attention mechanism enhances the inspection accuracy by adaptively weighting the feature maps and thus improving the detection efficiency. Finally, the GSconv convolutional layer combines the SC module and the DSC module in a channel shuffling manner to infiltrate the information of the SC module within some of the information of the DSC module, which improves the detection speed without changing the ***5s-PSG model outperforms the conventional target inspection in terms of accuracy, recall, map value, and loss function, where the accuracy is improved to 93.1%, recall to 91.2%, and map value to 94.6%. These findings are important for helmet inspection to find the optimal solution in terms of accuracy and speed for better helmet inspection in shop-floor factory environments.

关键词： Safety Head Accuracy Feature extraction YOLO Inspection Production facilities Attention mechanisms computational modeling Detection algorithms Helmet Yolov5s target detection attention mechanism

来源：评论

学校读者我要写书评

暂无评论

Integrating Data Collection, Communication, and Computation for Importance-Aware Online Edge Learning Tasks

引用

IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS 2025年第3期24卷 2606-2619页

作者： Wang, Nan Teng, Yinglei Huang, Kaibin Beijing Univ Posts & Telecommun BUPT Beijing Key Lab Work Safety Intelligent Monitoring Beijing 100876 Peoples R China Univ Hong Kong Dept Elect & Elect Engn Hong Kong Peoples R China

With the prevalence of real-time intelligence applications, online edge learning (OEL) has gained increasing attentions due to the ability of rapidly accessing environmental data to improve artificial intelligence models by edge computing. However, the performance of OEL is intricately tied to the dynamic nature of incoming data in ever-changing environments, which does not conform to a stationary distribution. In this work, we develop a data importance-aware collection, communication, and computation integration framework to boost the training efficiency by leveraging the varying data usefulness under dynamic network resources. A model convergence metric (MCM) is firstly derived that quantifies the data importance in mini-batch gradient descent (MGD)-based online learning tasks. To expedite model learning at the edge, we optimize training batch configuration and fine-tune the acquisition of important data through coordinated scheduling, encompassing data sampling, transmission and computational resource allocation. To cope with the time discrepancy and complex coupling of decision variables, we design a two-timescale hierarchical reinforcement learning (TTHRL) algorithm decomposing the original problem into two-layer subproblems and separately optimize the subproblems in a mixed timescale pattern. Experiments show that the proposed data integration framework can effectively improve the online learning efficiency while stabilizing caching queues in the system.

关键词： Data models computational modeling Training Data collection Optimization Wireless communication Convergence Resource management Measurement Data integration online learning two-timescale stochastic optimization hierarchical reinforcement learning queue stability

来源：评论

学校读者我要写书评

暂无评论

Budget-Constrained Resource Allocation and Pricing in VEC: A MSMLMF Stackelberg Game With Contract Incentive Mechanism

引用

IEEE INTERNET OF THINGS JOURNAL 2025年第5期12卷 5050-5067页

作者： Chen, Yishan Wu, Jie Ye, Shumei Li, Wei Xu, Zhonghui Jiangxi Univ Sci & Technol Coll Informat Engn Ganzhou 341000 Peoples R China Jiangxi Univ Sci & Technol Sch Sci Ganzhou 341000 Peoples R China

As a rapid increase in Internet of Things (IoT) devices, vehicle fog/edge computing (VFC/VEC) has seen swift development. Consequently, these devices often opt to purchase computing resources from the edge-cloud service providers (ECSPs) to expand their capabilities. However, due to limited edge server resources, they may face risks of overload or breakdowns. Meanwhile, there are often large amounts of underutilized idle resources near roads (such as parked vehicles), which can provide additional computing and communication capabilities to the system. Inspired by this, we propose a scheme that utilizes idle vehicles to assist in computation. To coordinate the interests of various participating entities and incentivize resource sharing, we construct a multistage multileader multifollower (MSMLMF) Stackelberg game model that encompasses the collaboration and competition among users, ECSPs, vehicle operators (VOPs), and idle vehicles. Participants aim to maximize their utility while considering the potential actions of others. Additionally, considering the information asymmetry between VOP and vehicles, we introduce individual rationality (IR) and incentive compatibility (IC) constraints from the contract theory to analyze and ensure the effectiveness of contracts. Next, we employ backward induction to gradually simplify the game model into convex optimization problems and theoretically prove the existence and uniqueness of Nash equilibrium (NE) points. Finally, through simulation experiments verify that our proposed model and scheme outperform other baselines in overall social welfare.

关键词： Resource management Games Contracts Servers Internet of Things computational modeling Security Scalability Pricing computational efficiency Contract theory idle resources incentive mechanism Internet of Things (IoT) lagrangian dual pricing stackelberg game

来源：评论

学校读者我要写书评

暂无评论

Federated Self-Supervised Learning Based on Prototypes Clustering Contrastive Learning for Internet of Vehicles Applications

引用

IEEE INTERNET OF THINGS JOURNAL 2025年第5期12卷 4692-4700页

作者： Dai, Cheng Wei, Shuai Dai, Shengxin Garg, Sahil Kaddoum, Georges Hossain, M. Shamim Sichuan Univ Sch Comp Sci Chengdu 610042 Peoples R China Ecole Technol Super Elect Engn Dept Montreal PQ H3C 1K3 Canada Chitkara Univ Inst Engn & Technol Ctr Res Impact & Outcome Rajpura 140401 India Lebanese Amer Univ Artificial Intelligence & Cyber Syst Res Ctr Beirut 03797751 Lebanon King Saud Univ Coll Comp & Informat Sci Dept Software Engn Riyadh 12372 Saudi Arabia

Federated learning (FL) is a novel paradigm for distribute edge intelligence for the Internet-of-Vehicles (IoV) application, which can enable superior performance in model training without the need to share local data. However, in the actual architecture of FL, the existence of nonindependent and identically distributed (non-IID) data at the edge device, along with the involvement of randomly participating distributed nodes, can result in model bias and a subsequent decrease in overall performance. To solve this problem, a new federated self-supervised learning method based on prototypes clustering contrastive learning (FedPCC) is proposed, which can effectively addresses the issue of asynchronous edge training and global model bias by introducing an unsupervised prototypes layer. The prototypes layer maps edge features to a global space and performs clustering, facilitating the new aggregation method of global prototypes on the server. Then, models from other components are aggregated based on data weight. Besides that, during the parameter deployment phase, we replace the prototype layer to acquire global knowledge, while employing momentum updates to preserve the local knowledge of the other components. Finally, to assess the efficacy of our proposed approach, we carried out comprehensive experiments across the various data sets. The findings show that our method gains state-of-the-art performance, which also validates its effectiveness.

关键词： Prototypes Servers Data models Training computational modeling Biological system modeling Contrastive learning federated learning (FL) Internet of Things (IoT) self-supervised learning

来源：评论

学校读者我要写书评

暂无评论

Interpretable Switching Deep Markov Model for Industrial Process Monitoring via Cloud-Edge Collaborative Framework

引用

IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT 2025年 74卷

作者： Sui, Qingkai Wang, Yalin Liu, Chenliang Huang, Keke Yang, Chunhua Cent South Univ Sch Automat Changsha 410083 Peoples R China

In modern manufacturing, the stability and safety of production processes. However, the frequent changes in industrial conditions necessitate timely updates and retraining of the on-site deployment of data-driven process monitoring methods, which is a task unattainable with the limited computational resources of edge devices. To address this issue, this article proposes an interpretable switching deep Markov model (ISDMM)-based cloud-edge collaborative framework for industrial process monitoring. ISDMM defines discrete switching variables that represent working conditions and corresponding multiple transition networks. The transition networks are concurrently trained and switched based on the current working condition, enabling ISDMM to capture different dynamics of the system under different conditions. Before deployment, model simplification is performed to construct a model base, where each model is tailored to individual working conditions. Moreover, in the edge layer, a model update strategy is designed to determine whether the model can be invoked from the model library based on the current working condition, alleviating the burden of unnecessary retraining. Finally, the effectiveness of ISDMM and its cloud-edge collaborative framework is validated through a numerical example and a real-world industrial process. industrial process.

关键词： computational modeling Data models Switches Collaboration Process monitoring Training Employee welfare Cloud computing Image edge detection Adaptation models Cloud-edge collaboration deep Markov model (DMM) model update and simplification process monitoring switching variables

来源：评论

学校读者我要写书评

暂无评论

GlucoseHD: Predicting Glucose Levels Using Hyperdimensional Computing

IEEE DESIGN & TEST

引用

IEEE DESIGN & TEST 2025年第2期42卷 17-24页

作者： Ponzina, Flavio Gomez, Mialyssa Xu, Congge Rosing, Tajana Simunic Univ Calif San Diego Syst Energy Efficiency Lab La Jolla CA 92093 USA

Editor’s notes: This article proposes an efficient algorithm to predict glucose levels using hyperdimensional computing. Highly accurate glucose level (GL) predictions are of critical importance for people affected by type 1 diabetes. To address this, the article proposes GlucoseHD, a hyperdimensional computing (HDC)-based approach for accurate GL predictions with limited computing and memory budgets. —Ganapati Bhat, Washington State University, USA Pietro Mercati, Intel Labs, USA

关键词： Glucose Predictive models computational modeling Training Encoding Data models Accuracy Hyperdimensional Computing Control Glucose Monitoring Glucose Level Predictions

来源：评论

学校读者我要写书评

暂无评论

Practical Deployment for Deep Learning-Based CSI Feedback Systems: Generalization Challenges and Enabling Techniques

引用

IEEE COMMUNICATIONS MAGAZINE 2025年

作者： Wang, Jintao Zhang, Xudong Lu, Zhilin Tan, Jingbo Zhang, Hengyu Tsinghua Univ Beijing Peoples R China Tsinghua Univ Dept Elect Engn Beijing Peoples R China Beijing Natl Res Ctr Informat Sci & Technol Beijing Peoples R China

For massive multiple-input multiple-output (MIMO) systems, downlink channel state information (CSI) feedback is critical for exploiting the performance potentials. Recognized by the 3rd Generation Partnership Project as a key use case for the artificial intelligence-native air interface, deep learning (DL)-based CSI feedback has attracted extensive research from academia and industry. However, one of the most significant obstacles to its practical deployment is the limited generalization capabilities of DL models. In this article, we first illustrate the generalization challenge faced by DL-based CSI feedback with the main factors. Then, we systematically analyze its feasible solutions from two perspectives: initial model acquisition and efficient online updating. Finally, to address knowledge-forgetting problems in online updating, a novel alternating optimization framework is proposed to further improve the generalization ability through knowledge review. Hopefully, this article can give inspiring insights for the optimization of deployment and generalization for DL-based CSI feedback systems.

关键词： Training Data models Decoding computational modeling 3GPP Optimization Adaptation models Accuracy Switches Multitasking

来源：评论

学校读者我要写书评

暂无评论

Sampling-Based Model Predictive Control Leveraging Parallelizable Physics Simulations

引用

IEEE ROBOTICS AND AUTOMATION LETTERS 2025年第3期10卷 2750-2757页

作者： Pezzato, Corrado Salmi, Chadi Trevisan, Elia Spahn, Max Alonso-Mora, Javier Corbato, Carlos Hernandez Delft Univ Technol Cognit Robot Dept NL-2628 CD Delft Netherlands

We present a sampling-based model predictive control method that uses a generic physics simulator as the dynamical model. In particular, we propose a Model Predictive Path Integral controller (MPPI) that employs the GPU-parallelizable IsaacGym simulator to compute the forward dynamics of the robot and environment. Since the simulator implicitly defines the dynamic model, our method is readily extendable to different objects and robots, allowing one to solve complex navigation and contact-rich tasks. We demonstrate the effectiveness of this method in several simulated and real-world settings, including mobile navigation with collision avoidance, non-prehensile manipulation, and whole-body control for high-dimensional configuration spaces. This is a powerful and accessible open-source tool to solve many contact-rich motion planning tasks.

关键词： Robots Collision avoidance Predictive models computational modeling Physics Trajectory Planning Cost function Real-time systems Optimal control Optimization and Optimal Control Contact modeling Whole-Body Motion Planning and Control Model Predictive Path Integral Control

来源：评论

学校读者我要写书评

暂无评论

SHAP-assisted EE-LightGBM model for explainable fault diagnosis in practical optical networks

引用

JOURNAL OF OPTICAL COMMUNICATIONS AND NETWORKING 2025年第2期17卷 81-94页

作者： Zhang, Chunyu Chen, Yu Zhang, Min Liu, Zhuo Wang, Danshi Beijing Univ Posts & Telecommun State Key Lab Informat Photon & Opt Commun Beijing 100876 Peoples R China China Mobile Commun Corp Beijing 100033 Peoples R China

Reliable fault diagnosis is crucial for ensuring the stable operation of optical networks. Recently, data-driven techniques have demonstrated significant advantages in fault diagnosis due to their outstanding data-processing capabilities and adaptive learning abilities. However, as equipment faults in practical optical networks are rare events, the data collected often faces severe data imbalance issues, greatly limiting the accuracy of traditional data-driven models. To address this challenge, a SHAP-assisted EE-LightGBM scheme is proposed for explainable fault diagnosis in practical optical networks. The EE-LightGBM model integrates undersampling strategies at the data level and hybrid ensemble strategies at the model level, enabling the full utilization of fewer fault samples and effectively alleviating the impact of data imbalance on model training. Furthermore, the SHAP method is used to explain the EE-LightGBM model. This method quantifies the contributions of input features to the model's decision outputs, facilitating a deeper understanding of the mechanisms underlying faults in the equipment and improving the model's explainability. Through SHAP analysis, we can determine key features highly correlated with equipment faults, thereby inferring the causes of equipment faults. Evaluation using data from backbone network equipment managed by operators shows excellent detection performance of the EE-LightGBM model at a data imbalance rate of 5.61%, with accuracy and F1 scores of 0.9968 and 0.9711, and false negative and false positive rates of 0.0033 and 0.0032, respectively. Moreover, the cause identification results are consistent with diagnostic expertise. We also explore the impact of data imbalance rates on the detection performance of the EE-LightGBM model. The model's low false negative rate under data imbalance further demonstrates its effectiveness in practical optical network fault diagnosis. (c) 2025 Optica Publishing Group

关键词： Data models Fault diagnosis Optical fiber networks computational modeling Feature extraction Training Fault detection Analytical models Decision making Limiting

来源：评论

学校读者我要写书评

暂无评论

Generative Artificial Intelligence-Oriented Synthetic Network: Toward Integrated Fine-Tuning and Inference When Generative Artificial Intelligence Meets Edge Intelligence in the Intelligent Internet of Vehicles

引用

IEEE VEHICULAR TECHNOLOGY MAGAZINE 2025年第2期20卷 83-94页

作者： Chen, Ning Cheng, Zhipeng Fan, Xuwei Liu, Zhang Yang, Jie Huang, Bangzhen Zhao, Yifeng Huang, Lianfen Du, Xiaojiang China Univ Petr East China Coll Control Sci & Engn Shandong Prov Engn Res Ctr Intelligent Sensing & M Qingdao 266580 Peoples R China Soochow Univ Sch Future Sci & Engn Suzhou 215006 Peoples R China Xiamen Univ Sch Informat Dept Commun Engn Xiamen 361000 Peoples R China Xiamen Univ Sch Informat & Commun Engn Xiamen 361000 Peoples R China Xiamen Univ Xiamen 361000 Peoples R China Xiamen Univ Sch Informat Dept Informat & Commun Engn Xiamen 361000 Peoples R China Xiamen Univ Dept Informat & Commun Engn Xiamen 361005 Peoples R China Xiamen Univ Dept Commun Engn Xiamen Peoples R China Stevens Inst Technol Dept Elect & Comp Engn Hoboken NJ 07030 USA

Generative artificial intelligence (GAI) and edge intelligence (EI) are driving the evolution of traditional vehicular networks toward the intelligent Internet of Vehicles (IIoV) by providing foundational and personalized knowledge. However, their inherently contradictory characteristics present significant challenges to direct knowledge sharing. Furthermore, conventional methods that independently optimize fine-tuning and inference lack the foresight to achieve long-term network benefits. To address these challenges, we propose the GAI-oriented synthetic network (GaisNet), a collaborative cloud-edge-end intelligence framework that integrates fine-tuning and inference. GaisNet, specifically, can mitigate contradictions by leveraging data-free knowledge relays, where bidirectional knowledge flow facilitates a virtuous cycle of model fine-tuning and task inference with a long-term perspective, fostering mutualism between GAI and EI in the IIoV. A case study illustrates the effectiveness of the proposed mechanisms. Finally, we discuss the future challenges and directions in the interplay between GAI and EI.

关键词： computational modeling Training Data models Artificial intelligence Transformers Frequency modulation Cloud computing Distributed databases Collaboration Pipelines

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：