检索结果-内蒙古大学图书馆

Efficient Federated Learning Using Dynamic Update and Adaptive Pruning with Momentum on Shared Server Data

ACM Transactions on Intelligent systems and technology 2024年第6期15卷 1-28页

作者： Liu, Ji Jia, Juncheng Zhang, Hong Yun, Yuhui Wang, Leye Zhou, Yang Dai, Huaiyu Dou, Dejing Hithink RoyalFlush Information Network Co. Ltd. Hangzhou China School of Computer Science and Technology Soochow University Suzhou China Collaborative Innovation Center of Novel Software Technology and Industrialization Nanjing China Baidu Research Beijing China Key Lab of High Confidence Software Technologies Ministry of Education and Software Institute Peking University Beijing China Department of Computer Science and Software Engineering Auburn University AuburnAL United States Department of Electrical and Computer Engineering North Carolina State University RaleighNC United States Bedi Cloud and School of Computer Science Fudan University Beijing China

Despite achieving remarkable performance, Federated Learning (FL) encounters two important problems, i.e., low training efficiency and limited computational resources. In this article, we propose a new FL framework, i.e., FedDUMAP, with three original contributions, to leverage the shared insensitive data on the server in addition to the distributed data in edge devices so as to efficiently train a global model. First, we propose a simple dynamic server update algorithm, which takes advantage of the shared insensitive data on the server while dynamically adjusting the update steps on the server in order to speed up the convergence and improve the accuracy. Second, we propose an adaptive optimization method with the dynamic server update algorithm to exploit the global momentum on the server and each local device for superior accuracy. Third, we develop a layer-adaptive model pruning method to carry out specific pruning operations, which is adapted to the diverse features of each layer so as to attain an excellent tradeoff between effectiveness and efficiency. Our proposed FL model, FedDUMAP, combines the three original techniques and has a significantly better performance compared with baseline approaches in terms of efficiency (up to 16.9 times faster), accuracy (up to 20.4% higher), and computational cost (up to 62.6% smaller). © 2024 Copyright held by the owner/author(s). Publication rights licensed to ACM.

关键词： Federated learning

来源：评论

学校读者我要写书评

暂无评论

CiC-NET: a real-time semantic segmentation network for dam surface crack detection

引用

Multimedia Tools and Applications 2024年 1-23页

作者： Li, Linjing Zhao, Hao Liu, Ran Nayyar, Anand Ali, Rashid Li, Yonglong Zhang, Hua School of Information Engineering Southwest University of Science and Technology Mianyang China University of Science and Technology of China Heifei China School of Electrical and Electronic Engineering Nanyang Technological University 50 Nanyang Avenue Singapore639798 Singapore School of Computer Science Duy Tan University Da Nang550000 Viet Nam Department of Computer Science University of Turbat Turbat Pakistan Department of Electronic Engineering Tsinghua University Beijing China Tianfu Institute of Research and Innovation Southwest University of Science and Technology Chengdu China

Crack detection is vital for maintaining hydraulic engineering infrastructure. However, achieving a balance between real-time processing and high precision in semantic segmentation models presents a significant challenge, especially given the intricate details of cracks and complex backgrounds. To tackle this issue, this paper proposes a real-time, high-precision crack segmentation model. Initially, a four-branch feature extraction structure is devised to capture the edge details of cracks, with a focus on enhancing segmentation accuracy. Subsequently, an image pyramid is constructed at the input end to feed small-scale samples into high-dimensional feature extraction branches, thereby reducing computational costs and improving segmentation speed. Finally, an effective feature fusion module is designed for the feature extraction structure to capture sufficient crack features and achieve precise crack localization. Extensive experiments validate the superior performance of the proposed method, with Pixel Accuracy, Recall, Intersection over Union, and F1 score reaching 68.45%, 67.04%, 51.22%, and 67.74%, respectively, while maintaining a modest parameter count of 13.12M. This model provides a reliable and efficient solution for the health monitoring and maintenance of hydraulic engineering. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2024.

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

Two-Stage Resource Scheduling for Deterministic Communication and Computation Integration

Two-Stage Resource Scheduling for Deterministic Communicatio...

引用

2024 IEEE Global Communications Conference, GLOBECOM 2024

作者： Zhang, Weiting Tang, Nian Zhang, Chuan Guo, Ruibin Li, Mingyan Ying, Chenhao Jin, Jian Beijing Jiaotong University School of Electronic and Information Engineering Beijing100044 China Beijing Institute of Technology School of Cyberspace Science and Technology Beijing100081 China Chongqing University College of Computer Science Chongqing400044 China Shanghai Jiao Tong University Department of Computer Science Shanghai200030 China Caict Research Institute of Industrial Internet of Things Beijing100095 China

ISBN: (纸本)9798350351255

In this paper, we investigate a resource orchestration and transmission scheduling problem for data-intensive services with diversified service requirements. A three-layer collaborative architecture is presented to support dynamic networking and computing resource allocation. To obtain optimal orchestration and scheduling policies, we formulate a constrained resource scheduling problem with the objective to maximizing resource utilization and scheduling success ratio. Since the complicated coupled constraints among decisions, we decouple the problem into a two-stage sub-problems of resource orchestration and transmission scheduling. To realize cross-domain resource orchestration and deterministic transmission of large-scale computing tasks, a two-stage resource scheduling scheme is proposed. Specifically, the first stage makes the resource orchestration decision by a greedy algorithm, and the second stage makes the transmission scheduling decision based on a deep reinforcement learning algorithm. Simulation results show that the proposed solution can effectively improve resource utilization and scheduling success ratio while satisfying diversified service requirements, as compared with benchmarks. © 2024 IEEE.

关键词： Deep reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Recent Advances in Space-Deployable Structures in China

引用

engineering 2022年第10期17卷 207-219页

作者： Xiaofei Ma Tuanjie Li Jingya Ma Zhiyi Wang Chuang Shi Shikun Zheng Qifeng Cui Xiao Li Fan Liu Hongwei Guo Liwu Liu Zuowei Wang Yang Li Xi’an Institute of Space Radio Technology Xi’an 710100China School of Electromechanical Engineering Xidian UniversityXi’an 710071China Beijing Institute of Spacecraft System Engineering Beijing 100094China Aerospace System Engineering Shanghai Shanghai 201109China State Key Laboratory of Robotics Harbin Institute of TechnologyHarbin 150001China Department of Astronautical Science and Mechanics Harbin Institute of TechnologyHarbin 150001China

Deployable space structure technology is an approach used in building spacecraft,especially when realizing deployment and folding *** in orbit,the structures are released from the fairing,deployed,and *** the development of communication,remote-sensing,and navigation satellites,space-deployable structures have become cutting-edge research topics in space science and *** paper summarizes the current research status and development trend of spacedeployable structures in China,including large space mesh antennas,space solar arrays,and deployable structures and mechanisms for deep-space *** technologies of space-deployable structures are addressed from the perspectives of deployable mechanisms,cable-membrane form-finding,dynamic analysis,reliable environmental adaptability analysis,and ***,future technology developments and trends are elucidated in the fields of mesh antennas,solar arrays,deployable mechanisms,and on-orbit adjustment,assembly,and construction.

关键词： Deployable space structures Mesh antennas Solar arrays Deep-space exploration

来源：评论

学校读者我要写书评

暂无评论

YOLOv7-Based Land and Underwater Target Detection and Recognition 20

YOLOv7-Based Land and Underwater Target Detection and Recogn...

引用

20th IEEE International Conference on Mechatronics and Automation, ICMA 2023

作者： Li, Jieren Shi, Liwei Guo, Shuxiang Beijing Institute of Technology School of Medical Technology Beijing100081 China Beijing Institute of Technology School of Life Science Beijing100081 China Ministry of Industry and Information Technology Key Laboratory of Convergence Medical Engineering System and Healthcare Technology Beijing Institute of Technology Beijing100081 China Southern University of Science and Technology The Department of Electronic and Electrical Engineering Guangdong Shenzhen518055 China

ISBN: (纸本)9798350320831

There is a growing demand for marine resource development. Bionic amphibious robots can replace humans to conduct land and underwater exploration, which has important research significance. Deep learning has developed rapidly in recent years, and many kinds of target detection algorithms have emerged. We select one two-stage target detection algorithm Faster R-CNN, three single-stage target detection algorithms SSD, Centernet, and YOLOv7. We use each of these four algorithms to train the VOC2007 dataset in a deep learning environment. After the training is completed, these four models are evaluated and predicted separately. We find an algorithm that is most suitable for the amphibious robot applicationYOLOv7. Finally, we use the YOLOv7 model to detect the underwater dataset, and the results prove that the model is promising for detecting small underwater targets. © 2023 IEEE.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

science communication with science fiction movies

引用

The Innovation 2024年第2期5卷 17-18页

作者： Yuanzhuo Wang Jingyuan Li Yuanhao Cui Department of Computer Science Beijing Technology and Business UniversityNo.33 Fucheng RoadHaidianBeijing 100048China Institute of Computing Technology Chinese Academy of SciencesNo.6 Kexueyuan South RoadHaidianBeijing 100190China School of System Design and Intelligent Manufacturing Southern University of Science and TechnologyNo.1088 Xueyuan AvenueShenzhenGuangdong 518055China

“It is not as easy as it looks,”Holden Thorp,science’s chief editor,commented at the end of 2021,when explaining new research findings of the COVID-19 pandemic to the public had encountered great difficulties.1 As a matter of fact,science communication has never been easy because serious scientific knowledge is prone to be boring rather than *** is only one atypical medium for science communication,i.e.,the science fiction movie(sci-fi movie for short),that has a huge amount of fans.

关键词： communication editor movie

来源：评论

学校读者我要写书评

暂无评论

Dual-Recommendation Disentanglement Network for View Fuzz in Action Recognition

引用

IEEE Transactions on Image Processing 2023年 32卷 2719-2733页

作者： Liu, Wenxuan Zhong, Xian Zhou, Zhuo Jiang, Kui Wang, Zheng Lin, Chia-Wen Wuhan University of Technology School of Computer Science and Artificial Intelligence Wuhan430070 China Wuhan University of Technology Hubei Key Laboratory of Transportation Internet of Things School of Computer Science and Artificial Intelligence Wuhan430070 China Peking University School of Electronics Engineering and Computer Science Beijing100091 China Harbin Institute of Technology School of Computer Science and Technology Harbin150001 China Wuhan University School of Computer Science Wuhan430072 China National Tsing Hua University Department of Electrical Engineering Hsinchu30013 Taiwan

Multi-view action recognition aims to identify action categories from given clues. Existing studies ignore the negative influences of fuzzy views between view and action in disentangling, commonly arising the mistaken recognition results. To this end, we regard the observed image as the composition of the view and action components, and give full play to the advantages of multiple views via the adaptive cooperative representation among these two components, forming a Dual-Recommendation Disentanglement Network (DRDN) for multi-view action recognition. Specifically, 1) For the action, we leverage a multi-level Specific Information Recommendation (SIR) to enhance the interaction among intricate activities and views. SIR offers a more comprehensive representation of activities, measuring the trade-off between global and local information. 2) For the view, we utilize a Pyramid Dynamic Recommendation (PDR) to learn a complete and detailed global representation by transferring features from different views. It is explicitly restricted to resist the fuzzy noise influence, focusing on positive knowledge from other views. Our DRDN aims for complete action and view representation, where PDR directly guides action to disentangle with view features and SIR considers mutual exclusivity of view and action clues. Extensive experiments have indicated that the multi-view action recognition method DRDN we proposed achieves state-of-the-art performance over powerful competitors on several standard benchmarks. The code will be available at https://***/51cloud/DRDN. © 1992-2012 IEEE.

关键词： Economic and social effects

来源：评论

学校读者我要写书评

暂无评论

Design and Static Analysis of Feet-legs Device of a Natatores-like Amphibious Robot 20

Design and Static Analysis of Feet-legs Device of a Natatore...

引用

20th IEEE International Conference on Mechatronics and Automation, ICMA 2023

作者： Lin, Xinru Shi, Liwei Guo, Shuxiang Beijing Institute of Technology School of Medical Technology Beijing100081 China School of Life Science Beijing Institute of Technology Beijing100081 China Ministry of Industry and Information Technology Key Laboratory of Convergence Medical Engineering System and Healthcare Technology Beijing Institute of Technology Beijing100081 China Southern University of Science and Technology The Department of Electronic and Electrical Engineering Guangdong Shenzhen518055 China

ISBN: (纸本)9798350320831

With the increased demand for the exploration of the ocean, the focus of research in the field of robotics has changed. Amphibious robots show broad application prospects. In order to design an amphibious robot adapted to multiple environments, a feet-legs device inspired by natatores was designed. Each webbed foot is made up of three sets of four-linked rods that allow the toe bones to fold and shrink, enabling better imitation of the webbed feet motion of natatores paddling. In order to analyse and optimize the mechanical strength of the critical components of the device during operation, static analysis and structural improvements are made to the webbed feet and the roller of the cylindrical cam. The results show that the static strength of improved webbed feet and the cam roller meet the requirements. © 2023 IEEE.

关键词： Static analysis

来源：评论

学校读者我要写书评

暂无评论

Fatigue Analysis of Leg Structure of a Natatores-Like Amphibious Robot 20

Fatigue Analysis of Leg Structure of a Natatores-Like Amphib...

引用

20th IEEE International Conference on Mechatronics and Automation, ICMA 2023

作者： Niu, Yan Shi, Liwei Guo, Shuxiang Beijing Institute of Technology School of Medical Technology Beijing100081 China Beijing Institute of Technology School of Life Science Beijing100081 China Key Laboratory of Convergence Medical Engineering System and Healthcare Technology Beijing Institute of Technology Ministry of Industry and Information Technology Beijing100081 China Southern University of Science and Technology The Department of Electronic and Electrical Engineering Guangdong Shenzhen518055 China

ISBN: (纸本)9798350320831

With the increasing demand for ocean exploration and development, the research about the amphibious robot is of great significance. In this paper, a natatores-like amphibious robot was designed by using the driving method of paddling by webbed foot. This mode has the advantages of high propulsion efficiency, good mobility and good stability. Due to the characteristics of this propulsion mode, the leg structure of the robot is often applied with periodic loads. And therefore it is necessary to verify the mechanical performance of the leg structure. The elastoplastic finite element model of the leg structure of the robot was modelled, and the fatigue analysis of the model under the walking gait were carried out. The results showed that the fatigue failure of the thigh and the shank would not occur within the design life, which met the design requirements. The analysis results can provide reference for the material selection and structural optimization design of leg in the subsequent work. © 2023 IEEE.

关键词： Robots

来源：评论

学校读者我要写书评

暂无评论

Improved Deep Matrix Factorization for Signal Recovery in Random Frequency Diversity Array Radar 2

Improved Deep Matrix Factorization for Signal Recovery in Ra...

引用

2nd IEEE International Conference on Signal, Information and Data Processing, ICSIDP 2024

作者： Guo, Dayan Lin, Zijian Huang, Tianyao Qi, Yue He, Jie Zhang, Xiao-Ping Wang, Xueqian Li, Gang University of Science and Technology Beijing School of Computer & Communication Engineering Beijing China Tsinghua Shenzhen International Graduate School Institute of Data and Information Shenzhen China Tsinghua University Department of Electronic Engineering Beijing China

ISBN: (纸本)9798331515669

Motivated by the success of recent deep learning researches on radar, we consider a deep learning based matrix factorization method for suppressing the high range-angle side-lobes in random frequency diversity array (FDA) radars. While the existing deep matrix factorization approach is designed for real-valued situations and signals of FDA radars are complex-valued cases, we propose two methods to extend the existing approach into complex-valued framework: 1) adjusting the neural network structures and activation functions for complex-valued operations, and 2) constructing a real-valued representation of the original complex-valued input. Simulation results demonstrate that the proposed methods outperform traditional ones in multi-target scenarios. © 2024 IEEE.

关键词： Matrix factorization

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：