检索结果-内蒙古大学图书馆

distributed Networked Real-Time Learning

IEEE TRANSACTIONS ON CONTROL OF NETWORK SYSTEMS 2021年第1期8卷 28-38页

作者： Garcia, Alfredo Wang, Luochao Huang, Jeff Hong, Lingzhou Texas A&M Univ Dept Ind & Syst Engn College Stn TX 77843 USA Texas A&M Univ Dept Comp Sci & Engn College Stn TX 77843 USA

Many machine learning algorithms have been developed under the assumption that datasets are already available in batch form. Yet, in many application domains, data are only available sequentially overtime via compute nodes in different geographic locations. In this article, we consider the problem of learning a model when streaming data cannot be transferred to a single location in a timely fashion. In such cases, a distributed architecture for learning which relies on a network of interconnected "local" nodes is required. We propose a distributed scheme in which every local node implements stochastic gradient updates based upon a local data stream. To ensure robust estimation, a network regularization penalty is used to maintain a measure of cohesion in the ensemble of models. We show that the ensemble average approximates a stationary point and characterizes the degree to which individual models differ from the ensemble average. We compare the results with federated learning to conclude that the proposed approach is more robust to heterogeneity in data streams (data rates and estimation quality). We illustrate the results with an application to image classification with a deep learning model based upon convolutional neural networks.

关键词： Asynchronous computing distributed computing networks nonconvex optimization real-time machine learning

来源：评论

学校读者我要写书评

暂无评论

Concept for a distributed Picking Application Utilizing Robotics and Digital Twins 27

Concept for a Distributed Picking Application Utilizing Robo...

引用

IEEE 27th International Conference on Emerging Technologies and Factory Automation (ETFA)

作者： Riedlinger, Marc A. Tamizi, Mehran Ghafarian Tikekar, Juilee Redeker, Magnus Fraunhofer Inst Optron Syst Technol & Image Explo IOSB INA Lemgo Fraunhofer IOSB Lemgo Germany Univ Victoria Victoria BC Canada

ISBN: (数字)9781665499965

ISBN: (纸本)9781665499965

The following paper describes a concept for a grasping application which utilizes technologies from the fields of distributed computing, robotics and Digital Twins in the sense of Industry 4.0. Hereby, the goal of the application is to have a computer vision system detect toy bricks which a robot has to pick and pass to a user. The application is divided into loosely coupled and distributed services that communicate with one another using a message broker.

关键词： distributed computing Robotics Digital Twin

来源：评论

学校读者我要写书评

暂无评论

distributed quantile regression for massive heterogeneous data

引用

NEUROcomputing 2021年 448卷 249-262页

作者： Hu, Aijun Jiao, Yuling Liu, Yanyan Shi, Yueyong Wu, Yuanshan Huazhong Univ Sci & Technol Sch Math & Stat Wuhan 430074 Hubei Peoples R China Hubei Univ Educ Sch Math & Econ Wuhan 430205 Hubei Peoples R China Wuhan Univ Sch Math & Stat Wuhan 430072 Hubei Peoples R China China Univ Geosci Sch Econ & Management Wuhan 430074 Hubei Peoples R China Zhongnan Univ Econ & Law Sch Math & Stat Wuhan 430073 Hubei Peoples R China

Massive data sets pose great challenges to data analysis because of their heterogeneous data structure and limited computer memory. Jordan et al. (2019, Journal of American Statistical Association) has proposed a communication-efficient surrogate likelihood (CSL) method to solve distributed learning problems. However, their method cannot be directly applied to quantile regression because the loss function in quantile regression does not meet the smoothness requirement in CSL method. In this paper, we extend CSL method so that it is applicable to quantile regression problems. The key idea is to construct a surrogate loss function which relates to the local data only through subgradients of the loss function. The alternating direction method of multipliers (ADMM) algorithm is used to address computational issues caused by the non-smooth loss function. Our theoretical analysis establishes the consistency and asymptotic normality for the proposed method. Simulation studies and applications to real data show that our method works well. (c) 2021 Elsevier B.V. All rights reserved.

关键词： ADMM Communication-efficient distributed computing Massive data Quantile regression

来源：评论

学校读者我要写书评

暂无评论

On the Capacity of Secure distributed Batch Matrix Multiplication

引用

IEEE TRANSACTIONS ON INFORMATION THEORY 2021年第11期67卷 7420-7437页

作者： Jia, Zhuqing Jafar, Syed Ali Beijing Univ Posts & Telecommun Sch Artificial Intelligence Beijing 100876 Peoples R China Univ Calif Irvine Dept Elect Engn & Comp Sci Ctr Pervas Commun & Comp Irvine CA 92697 USA

The problem of secure distributed batch matrix multiplication (SDBMM) studies the communication efficiency of retrieving a sequence of desired matrix products AB = (A(1)B(1), A(2)B(2), ..., A(S)B(S)) from N distributed servers where the constituent matrices A = (A(1), A(2), ..., A(S)) and B = (B-1, B-2, ..., B-S) are stored in X-secure coded form, i.e., any group of up to X colluding servers learn nothing about A, B. It is assumed that A(s) is an element of F-q(LxK), B-s is an element of F-q(KxM), s is an element of {1, 2, ..., S} are uniformly and independently distributed and F-q is a large finite field. The rate of an SDBMM scheme is defined as the ratio of the number of bits of desired information that is retrieved, to the total number of bits downloaded on average. The supremum of achievable rates is called the capacity of SDBMM. In this work we explore the capacity of SDBMM, as well as several of its variants, e.g., where the user may already have either A or B available as side-information, and/or where the security constraint for either A or B may be relaxed. We obtain converse bounds, as well as achievable schemes for various cases of SDBMM, depending on the L, K, M, N, X parameters, and identify parameter regimes where these bounds match. In particular, the capacity for securely computing a batch of outer products of two vectors is (1- X/N)(+), for a batch of inner products of two (long) vectors the capacity approaches (1 - 2X/N)(+) as the length of the vectors approaches infinity, and in general for sufficiently large K (e.g., K > 2min(L, M)), the capacity C is bounded as (1 - 2X/N)(+) <= C < (1 - X/N)(+). A remarkable aspect of our upper bounds is a connection between SDBMM and a form of private information retrieval (PIR) problem, known as multi-message X-secure T -private information retrieval (MM-XSTPIR). Notable features of our achievable schemes include the use of cross-subspace alignment and a transformation argument that converts a scalar multipli

关键词： Capacity security distributed computing matrix multiplication

来源：评论

学校读者我要写书评

暂无评论

ADTT: A Highly Efficient distributed Tensor-Train Decomposition Method for IIoT Big Data

引用

IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS 2021年第3期17卷 1573-1582页

作者： Wang, Xiaokang Yang, Laurence T. Wang, Yihao Ren, Lei Deen, M. Jamal Univ Elect Sci & Technol China Sch Informat & Commun Engn Chengdu 611731 Peoples R China St Francis Xavier Univ Dept Comp Sci Antigonish NS B2G 2W5 Canada Sankuai Online Technol Co Ltd Beijing 100089 Peoples R China Beihang Univ Sch Automat Sci & Elect Engn Beijing 100191 Peoples R China Beihang Univ Beijing Adv Innovat Ctr Big Data Based Precis Med Beijing 100191 Peoples R China McMaster Univ Dept Elect & Comp Engn Hamilton ON L8S 4K1 Canada

The industrial Internet of Things (IIoT) is growing quickly due to increasing deployment and integration of smart sensors, instruments, and devices, and software using wired or wireless networks. Through this integrated hardware-software approach, industrial practices will improve significantly, resulting in industrial intelligence for more efficient manufacturing. To realize such industrial intelligence, significant developments in IIoT big data processing and analysis are required to uncover and use hidden essential and valuable information of the production process. But large-scale, streaming, multiattribute IIoT data from production processes are noisy and have redundancies. Therefore, a suitable data processing technique such as tensor-train that can handle these IIoT data is needed. However, existing tensor-train decomposition methods are inefficient and cannot meet the processing demands of the large-scale IIoT big data. In this article, we propose an advanced (improved and highly efficient) distributed tensor-train (ADTT) decomposition method with its incremental computational method for processing IIoT big data. Finally, experiments are carried out on a typical and publicly available IIoT dataset-the bearing test data to verify and measure the performances of the proposed ADTT method.

关键词： Tensors Big Data Matrix decomposition Production Intelligent sensors Big data distributed computing high-performance computing industrial Internet of Things (IIoT) tensor-train

来源：评论

学校读者我要写书评

暂无评论

A fast and fully distributed method for region-based image segmentation Fast distributed region-based image segmentation

引用

JOURNAL OF REAL-TIME IMAGE PROCESSING 2021年第3期18卷 793-806页

作者： Mazouzi, Smaine Guessoum, Zahia Skikda Univ 20 Aout 1955 Comp Sci Dept Skikda Algeria Univ Reims Sorbonne Univ CReSTIC Reims Lip6 Paris France

distributed and parallel computing techniques allow fast image processing, namely when these techniques are applied at the low and the medium level of a vision system. In this paper, a collective and distributed method for image segmentation is introduced and evaluated. The method is modeled as a multi-agent system, where the agents aim to collectively produce a region-based segmentation. Each agent starts searching for an acceptable region seed by randomly jumping within the image. Next, it performs a region growing around its position. Thus, several agents find themselves within the same homogeneous region and are organized in a graph where two agents are connected if they are within the same region. So, a unifying of the labels in a same region is collaboratively performed by the agents themselves. The proposed method was experimented on real range images from the ABW dataset and the Object Segmentation Database (OSD) one, and the obtained results were compared to those of some well-referenced methods from the literature. The evaluation results show that the proposed method provides fast and accurate image segmentation, allowing it to be deployed for real-time vision systems.

关键词： Image segmentation Region growing Pixel labeling distributed computing Real-time computer vision systems

来源：评论

学校读者我要写书评

暂无评论

Large-scale knowledge distillation with elastic heterogeneous computing resources

引用

CONCURRENCY AND COMPUTATION-PRACTICE & EXPERIENCE 2023年第26期35卷

作者： Liu, Ji Dong, Daxiang Wang, Xi Qin, An Li, Xingjian Valduriez, Patrick Dou, Dejing Yu, Dianhai Baidu Inc Beijing Peoples R China Univ Montpellier CNRS Inria LIRMM Montpellier France

Although more layers and more parameters generally improve the accuracy of the models, such big models generally have high computational complexity and require big memory, which exceed the capacity of small devices for inference and incurs long training time. In addition, it is difficult to afford long training time and inference time of big models even in high performance servers, as well. As an efficient approach to compress a large deep model (a teacher model) to a compact model (a student model), knowledge distillation emerges as a promising approach to deal with the big models. Existing knowledge distillation methods cannot exploit the elastic available computing resources and correspond to low efficiency. In this paper, we propose an Elastic Deep Learning framework for knowledge Distillation, that is, EDL-Dist. The advantages of EDL-Dist are threefold. First, the inference and the training process is separated. Second, elastic available computing resources can be utilized to improve the efficiency. Third, fault-tolerance of the training and inference processes is supported. We take extensive experimentation to show that the throughput of EDL-Dist is up to 3.125 times faster than the baseline method (online knowledge distillation) while the accuracy is similar or higher.

关键词： deep neural network distributed computing knowledge distillation

来源：评论

学校读者我要写书评

暂无评论

distributed Financial Exchanges: Security Challenges and Design Principles

引用

IEEE SECURITY & PRIVACY 2021年第1期19卷 54-64页

作者： Massacci, Fabio Chan Nam Ngo Univ Trento I-38123 Trento Italy Vrije Univ Amsterdam NL-1081 HV Amsterdam Netherlands

Implementing secure, distributed, and economically viable financial exchanges radically challenges traditional constructs such as zero knowledge and secure multiparty computation. To boost discussions of such practica... 详细信息

关键词： Cryptography Financial Data Processing Security Of Data distributed Futures Exchange Secure Multiparty Computation Zero Knowledge Economically Viable Financial Exchanges Design Principles Security Challenges distributed Financial Exchanges Contracts Protocols Privacy Bitcoin Economics Finance Computer Security distributed computing Financial Management

来源：评论

学校读者我要写书评

暂无评论

Towards effective continued pre-training of EU institutional LLMs on EuroHPC supercomputers 2

Towards effective continued pre-training of EU institutional...

引用

2nd EuroHPC user day, EuroHPC 2024

作者： Rausch, Ilja Bhaskar, Bhavani Safont-Andreu, Anna Ewetz, Hans Kolovratnik, David Oravecz, Csaba Runonen, Markus European Commission Directorate-General for Translation

Large Language Models (LLMs) are a significant advancement in artificial intelligence (AI), capable of learning from vast textual datasets and excelling in tasks such as text generation and translation. However, the current general LLMs often do not meet the specific requirements of the public sector and other entities in Europe due to various limitations, including in particular language coverage gaps. In response, the European Commission's Directorate-General for Translation (DGT), in the context of its partnership with the Directorate-General for Communications Networks, Content and Technology (DG CONNECT) under the Digital Europe programme, aims to leverage its high-quality multilingual data coming from all the European Union (EU) institutions to contribute to the European ecosystem of LLMs through continued pre-training of open-source models. This paper presents these ongoing efforts on the supercomputers provided by the European High Performance computing Joint Undertaking (EuroHPC JU), with a focus on adapting Meta 's open-weight LLMs to European linguistic diversity. To this end we leverage the datasets of the European Advanced Multilingual Information System (EURAMIS), a unique and voluminous corpus of multilingual text from all EU institutions. Our approach utilizes state-of-the-art AI tools, including Hugging Face libraries and DeepSpeed 's ZeRO-3 data parallelism. We report on the results of our experiments, including the human evaluation of our models and various automated benchmarks such as ARC and HellaSwag, and machine translation tasks. Our findings demonstrate the potential of continued pre-training for enhancing the multilingual capabilities of open source LLMs for Europe. © 2025 The Authors. Published by Elsevier B.V.

关键词： distributed computing generative artificial intelligence large language models

来源：评论

学校读者我要写书评

暂无评论

A distributed Power Transfer Limit Calculation Method for Multi-Area Interconnection Power Networks

引用

IEEE TRANSACTIONS ON POWER SYSTEMS 2021年第5期36卷 4723-4732页

作者： Dong, Xiaoming Hao, Xupeng Chen, Quan Wang, Chengfu Wang, Mengxia Shandong Univ Jinan 250061 Peoples R China

For multi-area interconnected power networks, each subarea may be operated by independent regulator, who oppose to open the access of private information such as network topology to the others. For this case, this study proposes a decomposition-coordination strategy to calculate power transfer limit (PTL) considering both saddle node bifurcation (SNB) and limit-induced bifurcation (LIB). The external coordination equations are derived based on the sensitivity analysis of boundary buses at tie-lines, to express the power exchange of subareas and the consistent boundary state variables. Power Flow (PF) results for each subarea system are separately calculated followed by the adjustment of tie-lines' PF injected into boundary buses according to the external coordination. Then the alternation between internal and external iterations, for which operators of subareas are not required to share private network information, could generate the same results as conventional PF that relies on the complete information. Moreover, an improved Continuation PF (CPF) model that includes the prediction of the PF of area tie-lines is proposed. The point of collapse (POC) method is employed to calculate PTL involving the SNB of the interconnected networks to improve the accuracy of the algorithm. Case studies demonstrate the effectiveness and reliability of the proposed method.

关键词： Mathematical model Power systems Power markets Interconnected systems Voltage measurement Topology Convergence Power system analysis computing multi-area interconnected power transfer limit distributed computing decomposition-coordination strategy

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：