检索结果-内蒙古大学图书馆

Coordinated Sum-Rate Maximization in Multicell MU-MIMO With Deep Unrolling

IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS 2023年第4期41卷 1120-1134页

作者： Schynol, Lukas Pesavento, Marius Tech Univ Darmstadt Commun Syst Grp D-64283 Darmstadt Germany

Coordinated weighted sum-rate maximization in multicell MIMO networks with intra- and intercell interference and local channel state at the base stations is recognized as an important yet difficult problem. A classical, locally optimal solution is obtained by the weighted minimum mean squared error (WMMSE) algorithm which facilitates a distributed implementation in multicell networks. However, it often suffers from slow convergence and therefore large communication overhead. To obtain more practical solutions, the unrolling/unfolding of traditional iterative algorithms gained significant attention. In this work, we demonstrate a complete unfolding of the WMMSE algorithm for transceiver design in multicell MU-MIMO interference channels with local channel state information. The resulting architecture termed GCN-WMMSE applies ideas from graph signal processing and is agnostic to different wireless network topologies, while exhibiting a low number of trainable parameters and high efficiency w.r.t. training data. It significantly reduces the number of required iterations while achieving performance similar to the WMMSE algorithm, alleviating the overhead in a distributed deployment. Additionally, we review previous architectures based on unrolling the WMMSE algorithm and compare them to GCN-WMMSE in their specific applicable domains.

关键词： Deep unrolling deep unfolding WMMSE algorithm coordinated downlink beamforming multiuser MIMO multicell network graph convolutional neural network

来源：评论

学校读者我要写书评

暂无评论

An Exhaustive Survey on the Methods and Applications of Graph neural networks 8th

An Exhaustive Survey on the Methods and Applications of Grap...

引用

8th International Conference on Information Science, Communication and Computing

作者： Fan, Dong Cheng, Jieren Tang, Xiangyan Lu, Kai Hainan Univ Sch Comp Sci & Technol Haikou 570228 Hainan Peoples R China Hainan Blockchain Technol Engn Res Ctr Haikou 570228 Hainan Peoples R China Sch Cyberspace Secur Sch Cryptol Haikou 570228 Hainan Peoples R China Hainan Vocat Coll Polit & Law Dept Publ Safety Technol Haikou 571100 Hainan Peoples R China

ISBN: (纸本)9789819617838;9789819617814;9789819617807

The widespread use of deep learning has achieved unprecedented success. neural networks have achieved remarkable results in natural language processing, image recognition and other application fields. The key to the success of neural networks lies in its efficient processing of Euclidean space data, but the results obtained when traditional neural networks are used in non-Euclidean spaces are often not satisfactory to researchers. Graph neural network has been widely developed as a neural network that can be used in non-Euclidean space. This paper starts with the basic knowledge of graphs, introduces graph neural networks, and makes a new classification of them. Then it illustrates its application from 18 application areas. Finally, I put forward my own views on the future development and challenges of graph neural networks.

关键词： Graph neural networks Convolution Recurrent Graph Generation Euclidean non-Euclidean

来源：评论

学校读者我要写书评

暂无评论

Edge-Federated Learning-Based Intelligent Intrusion Detection System for Heterogeneous Internet of Things

引用

IEEE ACCESS 2024年 12卷 81736-81757页

作者： Mahadik, Shalaka S. Pawar, Pranav M. Muthalagu, Raja Birla Inst Technol & Sci Pilani Dubai Campus Dubai U Arab Emirates

distributed denial of service (DDoS) is an awful cyber threat, becoming more prevalent with mature heterogeneous IoT (HetIoT) applications like intelligent agriculture, wearables, and self-driving cars. Developing intelligent intrusion detection systems (IDS) using deep learning (DL) techniques to protect HetIoT has sparked a lot of attention. Federated learning (FL) helps to train the IDS locally with local data while respecting data privacy. Edge computing (EC) enhances security by processing data closer to the edge network. Therefore, the research contributed by integrating EC, FL, and DL and the proposed Edge-FL-based IDS. The proposed Edge-FL-based IDS aims to enhance HetIoT security by safeguarding data privacy against DDoS attacks. The research developed a DL-based convolutional neural network (CNN) classifier and used the CICDDoS2019 dataset to evaluate the success rate of the proposed Edge-FL-based IDS against DDoS attacks. The research employed IID and non-IID data distributions with participant clients K = 3, K = 5, and K = 7. The findings indicate that the proposed Edge-FL-based IDS outperforms centralized and other state-of-the-art FL models. The proposed Edge-FL-based IDS correctly detects and classifies DDoS attacks with the following accuracy: (a) 8-class IID: K = 3 is 99.98%, K = 5 is 99.97%, K = 7 is 99.96%. (b) 8-class non-IID: K = 3 is 99.97%, K = 5 is 99.94%, K = 7 is 99.90%. (c) 12-class IID: K = 3 is 99.96%, K = 5 is 99.96%, K = 7 is 99.95%. (d)12-class non-IID: K = 3 is 99.07%, K = 5 is 96.44%, K = 7 is 52.39%.

关键词： Heterogeneous IoT distributed denial of service (DDoS) federated learning (FL) independent and identically distributed (IID) non-IID edge computing (EC)

来源：评论

学校读者我要写书评

暂无评论

Adaptive distributed Convolutional neural network Inference at the network Edge with ADCNN 20

Adaptive Distributed Convolutional Neural Network Inference ...

引用

49th International Conference on Parallel processing (ICPP)

作者： Zhang, Sai Qian Lin, Jieyu Zhang, Qi Harvard Univ Cambridge MA 02138 USA Univ Toronto Toronto ON Canada Microsoft Redmond WA USA

ISBN: (纸本)9781450388160

The emergence of the Internet of Things (IoT) has led to a remarkable increase in the volume of data generated at the network edge. In order to support real-time smart IoT applications, massive amounts of data generated from edge devices need to be processed using methods such as deep neural networks (DNNs) with low latency. To improve application performance and minimize resource cost, enterprises have begun to adopt Edge computing, a computation paradigm that advocates processing input data locally at the network edge. However, as edge nodes are often resource-constrained, running data-intensive DNN inference tasks on each individual edge node often incurs high latency, which seriously limits the practicality and effectiveness of this model. In this paper, we study the problem of distributed execution of inference tasks on edge clusters for Convolutional neural networks (CNNs), one of the most prominent models of DNN. Unlike previous work, we present Fully Decomposable Spatial Partition (FDSP), which naturally supports resource heterogeneity and dynamicity in edge computing environments. We then present a compression technique that further reduces network communication overhead. Our system, called ADCNN, provides up to 2.8x speed up compared to state-of-the-art approaches, while achieving a competitive inference accuracy.

关键词： Edge computing Convolutional neural networks distributed inference

来源：评论

学校读者我要写书评

暂无评论

METNet: A mesh exploring approach for segmenting 3D textured urban scenes

引用

ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING 2024年 218卷 498-509页

作者： Schreiber, Qendrim Wolpert, Nicola Schoemer, Elmar Stuttgart Univ Appl Sci Schellingstr 24 D-70174 Stuttgart Germany Johannes Gutenberg Univ Mainz Staudingerweg 9 D-55128 Mainz Germany

In this work, we present the neural network Mesh Exploring Tensor Net (METNet) for the segmentation of 3D urban scenes, that operates directly on textured meshes. Since triangular meshes have a very irregular structure, many existing approaches change the input by sampling evenly distributed point clouds on the meshes. The resulting simplified representation of the urban scenes has the advantage that state-of-the-art neural network architectures for point clouds can be used for the segmentation task. The disadvantages are that crucial geodesic information is lost and fora sufficiently good approximation of the scene many points are often necessary. Since memory on the GPU is limited, the consequence is that only small areas can be examined locally. To overcome these limitations, METNet generates its input directly from the textured triangular mesh. It applies anew star-shaped exploration strategy, starting from a triangle on the mesh and expanding in various directions. This way, a vertex based regular local tensor from the unstructured mesh is generated, which we call a Mesh Exploring Tensor (MET). By expanding on the mesh, a MET maintains the information about the connectivity and distance of vertices along the surface of the mesh. It also effectively captures the characteristics of large regions. Our new architecture, METNet is optimized for processing METs as input. The regular structure of the input allows METNet the use of established convolutional neural networks. Experimental results, conducted on two urban textured mesh benchmarks, demonstrate that METNet surpasses the performance of previous state-of-the-art techniques. METNet improves the previous state-of-theart results by 8.6% in terms of mean IoU (intersection over union) on the SUM dataset compared to the second-best method PSSNet, and by 3.2% in terms of mean F1 score on the H3D dataset compared to the second-best method ifp-SCN. Our source code is available at https://***/QenSchr/METNet.

关键词： Texture meshes Semantic segmentation Urban scene Large scene understanding Structured representation Artificial neural networks

来源：评论

学校读者我要写书评

暂无评论

A Dynamic Sliding Window Based Tensor Communication Scheduling Framework for distributed Deep Learning

引用

IEEE TRANSACTIONS ON network SCIENCE AND ENGINEERING 2025年第2期12卷 1080-1095页

作者： Gao, Yunqi Hu, Bing Mashhadi, Mahdi Boloursaz Wang, Wei Tafazolli, Rahim Debbah, Merouane Zhejiang Univ Hangzhou 310027 Peoples R China Univ Surrey 5GIC& 6G Inst Commun Syst ICS Guildford GU2 7XH England Khalifa Univ KU 6G Res Ctr Dept Comp & Informat Engn Abu Dhabi 127788 U Arab Emirates

Simultaneous tensor communication can effectively improve the scalability of distributed deep learning on large clusters. However, a fixed number of tensor blocks communicated concurrently violates the priority-based scheduling strategy and cannot minimize communication overheads. In this paper, we propose a novel simultaneous tensor communication framework, namely D-Credit, which transmits tensor blocks based on dynamic sliding windows to minimize per-iteration time in distributed DNN training. We build the mathematical model of D-Credit in two phases: (1) the overlap of gradient communication and backward propagation, and (2) the overlap of gradient communication and forward computation. We drive the optimal window sizes for the second phase analytically, and develop a greedy algorithm to efficiently determine the dynamic window sizes for the first phase of D-Credit. We implement the D-Credit architecture on PyTorch framework. Experimental results on two different GPU clusters demonstrate that at training speed, D-Credit can achieve up to 1.26x, 1.21x, 1.48x and 1.53x speedup compared to ByteScheduler, DeAR, PyTorch-DDP and WFBP, respectively. At energy consumption, D-Credit saves up to 17.8% and 25.1% of the training energy consumption compared to ByteScheduler and WFBP, respectively.

关键词： Tensors Training Processor scheduling Parallel processing Dynamic scheduling Computational modeling Artificial neural networks Computer architecture Mathematical models Energy consumption distributed deep learning data parallelism communication scheduling tensor partitioning generative pre-trained transformer (GPT)

来源：评论

学校读者我要写书评

暂无评论

UMPIPE: Unequal Microbatches-Based Pipeline Parallelism for Deep neural network Training

引用

IEEE TRANSACTIONS ON PARALLEL AND distributed SYSTEMS 2025年第2期36卷 293-307页

作者： Zhou, Guangyao Tian, Wenhong Buyya, Rajkumar Wu, Kui Southwest Jiaotong Univ Sch Comp & Artificial Intelligence Chengdu 610032 Sichuan Peoples R China Univ Elect Sci & Technol China Sch Informat & Software Engn Chengdu 610056 Sichuan Peoples R China Univ Melbourne Dept Comp & Informat Syst Cloud Comp & Distributed Syst CLOUDS Lab Melbourne Vic 3052 Australia Univ Victoria Dept Comp Sci Victoria BC V8P 5C2 Canada

The increasing need for large-scale deep neural networks (DNN) has made parallel training an area of intensive focus. One effective method, microbatch-based pipeline parallelism (notably GPipe), accelerates parallel training in various architectures. However, existing parallel training architectures normally use equal data partitioning (EDP), where each layer's process maintains identical microbatch-sizes. EDP may hinder training speed because different processes often require varying optimal microbatch-sizes. To address this, we introduce UMPIPE, a novel framework for unequal microbatches-based pipeline parallelism. UMPIPE enables unequal data partitions (UEDP) across processes to optimize resource utilization. We develop a recurrence formula to calculate the time cost in UMPIPE by considering both computation and communication processes. To further enhance UMPIPE's efficiency, we propose the Dual-Chromosome Genetic Algorithm for UMPIPE (DGAP) that accounts for the independent time costs of forward and backward propagation. Furthermore, we present TiDGAP, a two-level improvement on DGAP. TiDGAP accelerates the process by simultaneously calculating the end time for multiple individuals and microbatches using matrix operations. Our extensive experiments validate the dual-chromosome strategy's optimization benefits and TiDGAP's acceleration capabilities. TiDGAP can achieve better training schemes than baselines, such as the local greedy algorithm and the global greedy-based dynamic programming. Compared to (GPipe, PipeDream), UMPIPE achieves increases in training speed: (13.89,11.09)% for GPT1-14, (17.11,7.96)% for VGG16 and >=(170,100)% for simulation networks.

关键词： Training Parallel processing Pipelines Data models Computational modeling Genetic algorithms Costs Pipeline processing Tensors Transformers Parallel training data parallelism unequal data partitions pipeline parallelism genetic algorithm

来源：评论

学校读者我要写书评

暂无评论

A Lightweight Transformer Model using neural ODE for FPGAs

A Lightweight Transformer Model using Neural ODE for FPGAs

引用

37th IEEE International Parallel and distributed processing Symposium (IPDPS)

作者： Okubo, Ikumi Sugiura, Keisuke Kawakami, Hiroki Matsutani, Hiroki Keio Univ 3-14-1 HiyoshiKohoku Ku Yokohama 2238522 Japan

ISBN: (纸本)9798350311990

A transformer is an emerging neural network model that employs an attention mechanism. It has been adopted to various tasks and achieved a favorable accuracy compared to CNNs (Convolutional neural networks) and RNNs (Recurrent neural networks). Although the attention mechanism is recognized as a general-purpose component, many of the transformer models require a significant number of parameters and thus they are not suited to low-cost edge devices. Recently, a resource-efficient hybrid model that uses ResNet as a backbone architecture and replaces a part of its convolutional layers with an MHSA (Multi-Head Self-Attention) mechanism was proposed. In this paper, we significantly reduce the parameter size of this approach by using neural ODE as a backbone architecture for the MHSA mechanism. The proposed hybrid model reduces the parameter size by 97.3% compared to the original model without degrading the accuracy. Since the model size is quite small, it is implemented on Xilinx ZCU104 FPGA (Field Programmable Gate Array) board so that it can fully exploit on-chip BRAM/URAM resources. The FPGA implementation is evaluated in terms of resource utilization, accuracy, performance, and power consumption. The results demonstrate that it speeds up the model by up to 2.63 times compared to a software execution without accuracy degradation.

关键词： BoTNet CNN Multi Head Self Attention neural ODE Transformer Zynq

来源：评论

学校读者我要写书评

暂无评论

A Relaxed Energy Function Based Analog neural network Approach to Target Localization in distributed MIMO Radar

arXiv

引用

arXiv 2022年

作者： Zhao, Xiaoyu Li, Jun Guo, Qinghua The National Laboratory of Radar Signal Processing Xidian University Xi’an710071 China The School of Electrical Computer and Telecommunications Engineering University of Wollongong WollongongNSW2522 Australia

Analog neural networks are highly effective to solve some optimization problems, and they have been used for target localization in distributed multiple-input multiple-output (MIMO) radar. In this work, we design a new relaxed energy function based neural network (RNFNN) for target localization in distributed MIMO radar. We start with the maximum likelihood (ML) target localization with a complicated objective function, which can be transformed to a more tractable one with equality constraints by introducing some auxiliary variables. Different from the existing Lagrangian programming neural network (LPNN) methods, we further relax the optimization problem formulated for target localization, so that the Lagrangian multiplier terms are no longer needed, leading to a relaxed energy function with better convexity. Based on the relaxed energy function, a RNFNN is implemented with much simpler structure and faster convergence speed. Furthermore, the RNFNN method is extended to localization in the presence of transmitter and receiver location errors. It is shown that the performance of the proposed localization approach achieves the Cramér-Rao lower bound (CRLB) within a wider range of signal-to-noise ratios (SNRs). Extensive comparisons with the state-of-the-art approaches are provided, which demonstrate the advantages of the proposed approach in terms of performance improvement and computational complexity (or convergence speed). Copyright © 2022, The Authors. All rights reserved.

关键词： Signal to noise ratio

来源：评论

学校读者我要写书评

暂无评论

MODEL-distributed INFERENCE IN MULTI-SOURCE EDGE networkS

MODEL-DISTRIBUTED INFERENCE IN MULTI-SOURCE EDGE NETWORKS

引用

IEEE International Conference on Acoustics, Speech, and Signal processing (ICASSP)

作者： Li, Pengzhen Seferoglu, Hulya Koyuncu, Erdem Univ Illinois Dept Elect & Comp Engn Chicago IL USA

ISBN: (纸本)9798350302615

distributed inference techniques can be broadly classified into data-distributed and model-distributed schemes. In data-distributed inference (DDI), each worker carries the entire deep neural network (DNN) model, but processes only a subset of the data. However, feeding the data to workers results in high communication costs especially when the data is large. An emerging paradigm is model-distributed inference (MDI), where each worker carries only a subset of DNN layers. In MDI, a source device that has data processes a few layers of DNN and sends the output to a neighboring device. This process ends when all layers are processed in a distributed manner. In this paper, we investigate MDI with multiple sources, i.e., when more than one device has data. We design a multi-source MDI (MS-MDI), which optimizes task scheduling decisions across multiple source devices and workers. Experimental results on a real-life testbed of NVIDIA Jetson TX2 edge devices show that MS-MDI improves the inference time significantly as compared to baselines.

关键词： Model distribution model distributed inference deep neural network (DNN) distributed DNN

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：