检索结果-内蒙古大学图书馆

International Conference on Wireless communications and Signal Processing (WCSP)

作者： Xujie Li Jichang Shen Ying Sun Ziya Wang Xuedong Zheng College of Computer and Information Engineering Hohai University Nanjing China National Mobile Communications Research Laboratory Southeast University Nanjing China Key Lab for Wireless Sensor network and Communication Shanghai Institute of Microsystem and Information Technology Chinese Academy of Sciences Shanghai China Nanjing China-Spacenet Satellite Telecom Co.Ltd China

ISBN: (数字)9781728172361

ISBN: (纸本)9781728172378

For unmanned aerial vehicle (UAV)-assisted fog computing system, content caching is an effective way to reduce network delay. How to smartly sense the future requirement of the user equipment and cache the contents in advance is a big challenge. Meanwhile, the impact of cache content size and other related factors on cache performance are often overlooked, which results in low cache hit ratio. Therefore, this paper proposes a smart caching and caching replacement scheme based on valuation function of content caching for UAV-assisted fog computing network. Firstly, the effect of the content popularity, content size and available cache storage space on system performance is analysed. Then the valuation function combing these factors is defined. Based on the presented valuation function, a smart content caching and caching replacement scheme for UAV-assisted fog computing network is proposed. Finally, the simulation results validate the effectiveness of the proposed scheme. The scheme can be applied into the design the deployment of the UVA-assisted fog computing network.

关键词： Edge computing Cost accounting Unmanned aerial vehicles Performance evaluation Delays Indexes Quality of service

来源：评论

学校读者我要写书评

暂无评论

SiamIRPN: Siamese Visual Tracking with Improved Region Proposal networks

SiamIRPN: Siamese Visual Tracking with Improved Region Propo...

引用

Frontiers of Robotics and Software Engineering (FRSE), International Conference on

作者： Dun Cao Renhua Dai Tianyu Zhao Fayez Alqahtani Amr Tolba R. Simon Sherratt Min Zhu School of Computer and Communications Engineering Changsha University of Science and Technology Changsha Hunan China Key Lab of Broadcast Wireless Communication and Sensor Network Technology Nanjing University of Posts and Telecommunications Ministry of Education Software Engineering Department College of Computer and Information Sciences King Saud University Riyadh Saudi Arabia Computer Science Department Community College King Saud University Riyadh Saudi Arabia School of Biomedical Engineering The University of Reading United Kingdom College of Information Science and Technology Zhejiang Shuren University Zhejiang China

Though introducing the Region Proposal network (RPN) from object detection enabled Siamese trackers’ success, RPN-based trackers still struggle in challenging scenarios. We posit that the reason comes from two major limitations of the introduced RPN, where one is that the external structure of the RPN is simple and straight, the other is that the internal components used in RPN cannot cope with complex scenes. In this paper, we propose an Improved RPN (IRPN) suitable for visual tracking. Externally, we place a Convolutional Block Attention Mechanism (CBAM) to the IRPN, and internally, we adopt the advanced component groups for the IPRN. Using multiple IRPN blocks and deep architecture, we propose a Siamese tracker (SiamIRPN) having a layer-wise structure. Comprehensive experiments and ablation studies on five benchmarks (VOT-2019, OTB100, UAV123, GOT-10k, and NFS) show our proposed SiamIRPN achieves competitive performance.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Robust Generalization against Photon-Limited Corruptions via Worst-Case Sharpness Minimization

arXiv

引用

arXiv 2023年

作者： Huang, Zhuo Zhu, Miaoxi Xia, Xiaobo Shen, Li Yu, Jun Gong, Chen Han, Bo Du, Bo Liu, Tongliang Syndey Ai Centre The University of Sydney Australia National Engineering Research Center for Multimedia Software Institute of Artificial Intelligence School of Computer Science Hubei Key Laboratory of Multimedia and Network Communication Engineering Wuhan University China Jd Explore Academy China Department of Automation University of Science and Technology of China China Key Lab. of Intelligent Percept. and Syst. for High-Dimensional Information of Ministry of Education School of Computer Science and Engineering Nanjing University of Science and Technology China Department of Computer Science Hong Kong Baptist University Hong Kong

Robust generalization aims to tackle the most challenging data distributions which are rare in the training set and contain severe noises, i.e., photon-limited corruptions. Common solutions such as distributionally robust optimization (DRO) focus on the worst-case empirical risk to ensure low training error on the uncommon noisy distributions. However, due to the over-parameterized model being optimized on scarce worst-case data, DRO fails to produce a smooth loss landscape, thus struggling on generalizing well to the test set. Therefore, instead of focusing on the worst-case risk minimization, we propose SharpDRO by penalizing the sharpness of the worst-case distribution, which measures the loss changes around the neighbor of learning parameters. Through worst-case sharpness minimization, the proposed method successfully produces a flat loss curve on the corrupted distributions, thus achieving robust generalization. Moreover, by considering whether the distribution annotation is available, we apply SharpDRO to two problem settings and design a worst-case selection process for robust generalization. Theoretically, we show that SharpDRO has a great convergence guarantee. Experimentally, we simulate photon-limited corruptions using CIFAR10/100 and ImageNet30 datasets and show that SharpDRO exhibits a strong generalization ability against severe corruptions and exceeds well-known baseline methods with large performance gains. Copyright © 2023, The Authors. All rights reserved.

关键词： Optimization

来源：评论

学校读者我要写书评

暂无评论

Delay Optimization Based on Improved Differential Evolutionary Algorithm for Task Offloading in Fog Computing networks

Delay Optimization Based on Improved Differential Evolutiona...

引用

International Conference on Wireless communications and Signal Processing (WCSP)

作者： Xujie Li Guangzhao Zhang Xuedong Zheng Siyang Hua College of Computer and Information Hohai University Nanjing China National Mobile Communications Research Laboratory Southeast University Nanjing China Key Lab for Wireless Sensor network and Communication Chinese Academy of Sciences Shanghai China Nanjing China-Spacenet Satellite Telecom Co. Ltd Nanjing China Talent Science and Technology Co. Ltd Nanjing China

ISBN: (数字)9781728172361

ISBN: (纸本)9781728172378

With the rapid development of communication technology, more and more devices tend to be intelligent. Intelligent devices with computing functions can provide computing services to reduce total task completion time in fog computing networks. This paper aims at minimizing the total task completion time to study multi-user multi-node multi-task offloading problem. Firstly, the task offloading process of task node and fog node is modeled. Then, the system function of the total task completion time is constructed and an optimization problem with constraints is generated. The problem is NP-hard and difficult to be solved by traditional methods. Finally, the improved differential evolution (IDE) algorithm is used to solve the task offloading problem. Simulation results show that it has better performance and can significantly shorten the total task completion time compared with other algorithms.

关键词： Task analysis Delays Computational modeling Edge computing Optimization Signal to noise ratio Wireless communication

来源：评论

学校读者我要写书评

暂无评论

Outage Performance of Satellite-Aerial-Terrestrial network

Outage Performance of Satellite-Aerial-Terrestrial Network

引用

IEEE Globecom Workshops

作者： Qingquan Huang Min Lin Jun-Bo Wang Wei-Ping Zhu Communications Engineering College Army Engineering University of PLA Nanjing China The 63rd Research Institute National University of Defense Technology Nanjing China Key Lab of Broadband Wireless Communication and Sensor Network Technology Ministry of Education Nanjing University of Posts and Telecommunications Nanjing China National Mobile Communications Research Laboratory Southeast University Nanjing China Department of Electrical and Computer Engineering Concordia University Montreal Canada School of Communication and Information Engineering Nanjing University of Posts and Telecommunications Nanjing China

ISBN: (数字)9781728109602

ISBN: (纸本)9781728109619

In this paper, we investigate the performance of an amplify-and-forward satellite-aerial-terrestrial network (SATN) with a multi-antenna unmanned aerial vehicle (UAV) relay, where the satellite-UAV link undergoes correlated Shadowed-Rician fading while the UAV-terrestrial links experience correlated Rician fading, respectively. Specifically, we first formulate a constrained optimization problem in terms of maximizing the system capacity to obtain the beamforming (BF) weight vectors at the UAV. By assuming that perfect channel state information (CSI) is unavailable, we design statistical CSI-aided BF scheme to enhance the system performance. Then, with the help of probability theory, we obtain the closed-form expression of outage probability for the considered SATN. Simulation results are provided to validate the effectiveness of our proposed BF scheme and the correctness of the derived expressions.

关键词： Satellites Unmanned aerial vehicles Relays Antenna arrays Signal to noise ratio Wireless communication Optimization

来源：评论

学校读者我要写书评

暂无评论

MAC-AC: A Novel Distributed MAC Protocol for Accessing Channel in Vehicular Ad Hoc networks

MAC-AC: A Novel Distributed MAC Protocol for Accessing Chann...

引用

IEEE Conference on Vehicular technology (VTC)

作者： Baozhu Li Fen Hou Changyue Zhang Shaohua Li Shujuan Ji Shanzhi Chen School of Information Science and Engineering University of Jinan Jinan China Shandong Provincial Key Laboratory of Network-based Intelligent Computing Jinan China Faculty of Science and Technology University of Macau Macau China Automotive Research Institute China National Heavy Duty Truck Group Co. Ltd Jinan China Shandong Key Laboratory of Wisdom Mine Information Technology Qingdao China College of Computer Science and Engineering Shandong University of Science and Technology Qingdao China State Key Lab of Wireless Mobile Communication Beijing China China Academy of Telecommunication Technology Beijing China

ISBN: (数字)9781728194844

ISBN: (纸本)9781728194851

As a promising paradigm, VANET has been attracting more and more attention from the industry and academia recently. However, due to rapid movement of vehicles and highly dynamic topology, designing efficient MAC protocol for VANET is still challenging. In this paper, we propose MAC-AC, a novel TDMA-based distributed MAC protocol designed specifically for a vehicular ad hoc network. In MAC-AC, when multiple vehicles compete for the same time slot within their two-hop communication range, one of contending vehicle can obtain this time slot by a simple method, which increases the success probability of vehicles accessing channel. Analysis results are presented to demonstrate the efficiency of MAC-AC and compare it to ADHOC MAC, an existing MAC protocol based on TDMA.

关键词： Vehicular and wireless technologies Time division multiple access network topology Vehicular ad hoc networks Media Access Protocol Topology Vehicle dynamics

来源：评论

学校读者我要写书评

暂无评论

Pseudo-labeling Based Practical Semi-Supervised Meta-Training for Few-Shot Learning

arXiv

引用

arXiv 2022年

作者： Dong, Xingping Ouyang, Tianran Liao, Shengcai Du, Bo Shao, Ling School of Computer Science National Engineering Research Center for Multimedia Software Institute of Artificial Intelligence Hubei Key Laboratory of Multimedia and Network Communication Engineering Wuhan University Wuhan430072 China College of Information Technology United Arab Emirates University Al Ain United Arab Emirates UCAS-Terminus AI Lab University of Chinese Academy of Sciences Beijing100049 China

Most existing few-shot learning (FSL) methods require a large amount of labeled data in meta-training, which is a major limit. To reduce the requirement of labels, a semi-supervised meta-training (SSMT) setting has been proposed for FSL, which includes only a few labeled samples and numbers of unlabeled samples in base classes. However, existing methods under this setting require class-aware sample selection from the unlabeled set, which violates the assumption of unlabeled set. In this paper, we propose a practical semi-supervised meta-training setting with truly unlabeled data to facilitate the applications of FSL in realistic scenarios. To better utilize both the labeled and truly unlabeled data, we propose a simple and effective meta-training framework, called pseudo-labeling based meta-learning (PLML). Firstly, we train a classifier via common semi-supervised learning (SSL) and use it to obtain the pseudo-labels of unlabeled data. Then we build few-shot tasks from labeled and pseudo-labeled data and design a novel finetuning method with feature smoothing and noise suppression to better learn the FSL model from noise labels. Surprisingly, through extensive experiments across two FSL datasets, we find that this simple meta-training framework effectively prevents the performance degradation of various FSL models under limited labeled data, and also significantly outperforms the representative SSMT models. Besides, benefiting from meta-training, our method also improves several representative SSL algorithms as well. We provide the training code and usage examples at https://***/ouyangtianran/PLML. © 2022, CC BY-NC-ND.

关键词： Self-supervised learning

来源：评论

学校读者我要写书评

暂无评论

Understanding the Robustness of 3D Object Detection with Bird'View Representations in Autonomous Driving

Understanding the Robustness of 3D Object Detection with Bir...

引用

Conference on computer Vision and Pattern Recognition (CVPR)

作者： Zijian Zhu Yichi Zhang Hai Chen Yinpeng Dong Shu Zhao Wenbo Ding Jiachen Zhong Shibao Zheng Institute of Image Communication and Network Engineering Shanghai Jiao Tong University Dept. of Comp. Sci. and Tech. THBI Lab Institute for AI Tsinghua University BNRist Center Zhongguancun Laboratory Key Laboratory of Intelligent Computing and Signal Processing Ministry of Education Information Materials and Intelligent Sensing Laboratory of Anhui Province School of Computer Science and Technology Anhui University SAIC Motor AI Lab

3D object detection is an essential perception task in autonomous driving to understand the environments. The Bird's-Eye-View (BEV) representations have significantly improved the performance of 3D detectors with camera inputs on popular benchmarks. However, there still lacks a systematic understanding of the robustness of these vision-dependent BEV models, which is closely related to the safety of autonomous driving systems. In this paper, we evaluate the natural and adversarial robustness of various representative models under extensive settings, to fully understand their behaviors influenced by explicit BEV features compared with those without BEV. In addition to the classic settings, we propose a 3D consistent patch attack by applying adversarial patches in the 3D space to guarantee the spatiotemporal consistency, which is more realistic for the scenario of autonomous driving. With substantial experiments, we draw several findings: 1) BEV models tend to be more stable than previous methods under different natural conditions and common corruptions due to the expressive spatial representations; 2) BEV models are more vulnerable to adversarial noises, mainly caused by the redundant BEV features; 3) Camera-LiDARfusion models have superior performance under different settings with multi-modal inputs, but BEV fusion model is still vulnerable to adversarial noises of both point cloud and image. These findings alert the safety issue in the applications of BEV detectors and could facilitate the development of more robust models.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Are Dense labels Always Necessary for 3D Object Detection from Point Cloud?

arXiv

引用

arXiv 2024年

作者： Gao, Chenqiang Liu, Chuandong Shu, Jun Liu, Fangcen Liu, Jiang Yang, Luyu Gao, Xinbo Meng, Deyu The School of Intelligent Systems Engineering The Shen-zhen Campus of Sun Yatsen University Sun Yat-sen University Guangdong Shenzhen518107 China The School of Computer Science Wuhan University Wuhan430072 China School of Mathematics and Statistics Ministry of Education Key Lab of Intelligent Networks and Network Security Xi’an Jiaotong University Shaanxi 710049 China School of Communication and Information Engineering Chongqing University of Posts and Telecommunications Chongqing400065 China Meta Menlo Park94025 United States University of Maryland College Park20742 United States

Current state-of-the-art (SOTA) 3D object detection methods often require a large amount of 3D bounding box annotations for training. However, collecting such large-scale densely-supervised datasets is notoriously costly. To reduce the cumbersome data annotation process, we propose a novel sparsely-annotated framework, in which we just annotate one 3D object per scene. Such a sparse annotation strategy could significantly reduce the heavy annotation burden, while inexact and incomplete sparse supervision may severely deteriorate the detection performance. To address this issue, we develop the SS3D++ method that alternatively improves 3D detector training and confident fully-annotated scene generation in a unified learning scheme. Using sparse annotations as seeds, we progressively generate confident fully-annotated scenes based on designing a missing-annotated instance mining module and reliable background mining module. Our proposed method produces competitive results when compared with SOTA weakly-supervised methods using the same or even more annotation costs. Besides, compared with SOTA fully-supervised methods, we achieve on-par or even better performance on the KITTI dataset with about 5× less annotation cost, and 90% of their performance on the Waymo dataset with about 15× less annotation cost. The additional unlabeled training scenes could further boost the performance. The code will be available at https://***/gaocq/SS3D2. Copyright © 2024, The Authors. All rights reserved.

关键词： Object detection

来源：评论

学校读者我要写书评

暂无评论

ZF-Based Beamforming for Wireless Powered Cognitive Satellite-Terrestrial networks

ZF-Based Beamforming for Wireless Powered Cognitive Satellit...

引用

IEEE Global communications Conference

作者： Zhi Lin Min Lin Tomaso de Cola Benoit Champagne A. Lee Swindlehurst College of Communications Engineering Army Engineering University of PLA Nanjing China Key Lab of Broadband Wireless Communication and Sensor Network Technology Ministry of Education Nanjing University of Posts and Telecommunications Nanjing China Institute of Communications and Navigation German Aerospace Center (DLR) Germany Department of Electrical and Computer Engineering McGill University Montreal Canada Center for Pervasive Communications and Computing University of California Irvine Irvine USA

ISBN: (数字)9781728109626

ISBN: (纸本)9781728109633

In this paper, we propose a novel zero-forcing (ZF)-based beamforming (BF) scheme for a wireless powered cognitive satellite-terrestrial network (CSTN) operated in the millimeter wave band. Assuming that the satellite and base station are equipped with multiple antennas, we aim at maximizing the sum rate of the CSTN while satisfying the signal-to-interference-plus-noise-ratio requirements for both the information receivers (IRs) and earth stations, the energy harvesting requirements of the energy receivers (ERs), and the secrecy constraints at the ERs. Since the resulting optimization problem is mathematically intractable, we propose a novel multi-beam-based ZF BF scheme to generate beamforming vectors to serve the IRs and ERs. Specifically, the original nonconvex problem is decomposed into two independent subproblems. The first subproblem, which features beam orthogonality constraints, leads to closed form solutions for the beamforming vectors. The second subproblem, aiming at finding the optimal power allocation, is solved via the S-procedure. Finally, the effectiveness of the proposed scheme is demonstrated by simulation results.

关键词： Erbium Satellite broadcasting Wireless communication Satellite antennas Array signal processing Energy harvesting Optimization

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：