检索结果-内蒙古大学图书馆

IEEE International Conference on Software Engineering and Service Sciences (ICSESS)

作者： Penghui Ma Peichang Shi Guodong Yi National Key Laboratory of Parallel and Distributed Processing College of Computer Science National University of Defense Technology Changsha China Key Laboratory of Software Engineering for Complex Systems College of Computer Science National University of Defense Technology Changsha China Xiangjiang Lab Changsha China School of Advanced Interdisciplinary Studies Hunan University Of Technology and Business Changsha China

With serverless computing offering more efficient and cost-effective application deployment, the diversity of serverless platforms presents challenges to users, including platform lock-in and costly migration. Moreover, due to the black box nature of function computing, traditional performance benchmarking methods are not applicable, necessitating new studies. This article presents a detailed comparison of six major public cloud function computing platforms and introduces a benchmarking framework for function computing performance. This framework aims to help users make comprehensive comparisons and select the most suitable platform for their specific needs.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Trace Bench: An Open Data Set for Trace-Oriented Monitoring

Trace Bench: An Open Data Set for Trace-Oriented Monitoring

引用

IEEE International Conference on Cloud Computing Technology and Science (CloudCom)

作者： Jingwen Zhou Zhenbang Chen Ji Wang Zibin Zheng Michael R. Lyu College of Computer National University of Defense Technology Changsha China Science and Technology on Parallel and Distributed Processing Laboratory National University of Defense Technology Changsha China Shenzhen Research Institute The Chinese University of Hong Kong Shenzhen China Department of Computer Science and Engineering The Chinese University of Hong Kong Hong Kong China

User request trace-oriented monitoring is an effective method to improve the reliability of cloud systems. However, there are some difficulties in getting traces in practice, which hinder the development of trace-oriented monitoring research. In this paper, we release a fine-grained user request-centric open trace data set, called Trace Bench, collected on a real world cloud storage system deployed in a real environment. During collecting, many aspects are considered to simulate different scenarios, including cluster size, request type, workload speed, etc. Besides recording the traces when the monitored system is running normally, we also collect the traces under the situation with faults injected. With a mature injection tool, 14 faults are introduced, including function faults and performance faults. The traces in Trace Bench are clustered in different files, where each file corresponds to a certain scenario. The whole collection work lasted for more than half a year, resulting in more than 360, 000 traces in 361 files. In addition, we also employ several applications based on Trace Bench, which validate the helpfulness of Trace Bench for the field of trace-oriented monitoring.

关键词： Monitoring Servers Educational institutions Instruments Cloud computing Context Fault diagnosis

来源：评论

学校读者我要写书评

暂无评论

A GPU-based parallel WFST decoder on nnet3

引用

AIP Conference Proceedings 2019年第1期2073卷

作者： Yong Wang Jie Liu Chen Zhou Zhengbin Pang Shengguo Li Chunye Gong Xinbiao Gan Yurong Li 1Science and Technology on Parallel and Distributed Processing Laboratory National University of Defense Technology Changsha 410073 China

One performance-intensive part of automatic speech recognition is the weighted finite-state transducer (WFST) decoding. To solve the problem, we expand parallel Graphics processing Units (GPU) computing to the decoding period. We describe extension work based on Kaldi toolkit for speech recognition research. Our work can support weighted finite-state transducer decoding on Kaldi neural nets with CUDA toolkit. Our paper also expands an efficient parallel Viterbi beam decoding algorithm to decrease the speech recognition Real Time Factor (RTF) value. Together with our optimization algorithm, we have reached 2.3x speed up on the AISHELL corpus decoding. We also implement nnet3 decoder that improves real-time speed up with no word error rate raise.

关键词：

来源：评论

学校读者我要写书评

暂无评论

科学研究的第五范式——以智能驱动的材料设计为例

引用

Engineering 2023年第5期24卷 126-137,I0003,I0004页

作者： Can Leng Zhuo Tang Yi-Ge Zhou Zean Tian Wei-Qing Huang Jie Liu Keqin Li Kenli Li Science and Technology on Parallel and Distributed Processing Laboratory National University of Defense TechnologyChangsha 410073China Laboratory of Software Engineering for Complex Systems National University of Defense TechnologyChangsha 410073China National Supercomputing Center in Changsha Changsha 410082China College of Computer Science and Electronic Engineering Hunan UniversityChangsha 410082China Institute of Chemical Biology and Nanomedicine State Key Laboratory of Chemo/Biosensing and ChemometricsCollege of Chemistry and Chemical EngineeringHunan UniversityChangsha 410082China Department of Applied Physics School of Physics and ElectronicsHunan UniversityChangsha 410082China Department of Computer Science State University of New YorkNew PaltzNY 12561USA

科学正在进入一个新时代——第五范式——它被认为是知识整合到不同领域的主要特征,是基于无所不在的机器学习系统的计算社区中智能驱动的工作。在此,我们通过在天河一号超级计算机系统上构建的催化材料专门设计的典型平台案例,生动地... 详细信息

科学正在进入一个新时代——第五范式——它被认为是知识整合到不同领域的主要特征,是基于无所不在的机器学习系统的计算社区中智能驱动的工作。在此,我们通过在天河一号超级计算机系统上构建的催化材料专门设计的典型平台案例,生动地阐明了第五范式的本质,旨在促进第五范式在其他领域的培养。第五范式平台主要包括模型自动构建(原始数据提取)、指纹自动构建(神经网络特征选择)以及跨学科知识串联的重复迭代(“火山图”)。与分解一起进行的是对迭代中实现的体系结构的性能评估。通过讨论,第五范式的智能驱动平台可以极大地简化和改进研究中极其繁琐和具有挑战性的工作,并通过补偿机器学习中缺少样本和替代一些由于计算资源不足而导致的数值计算来实现数值计算与机器学习之间的相互反馈,从而加速探索过程。在数据驱动的学科中,跨学科专家的协同作用和对动态数据需求的急剧增长仍然是一个挑战。我们相信,对第五范式平台的一瞥可以为其在其他领域的应用铺平道路。

关键词：机器学习自动构建天河一号数据驱动神经网络动态数据知识整合科学研究

来源：评论

学校读者我要写书评

暂无评论

Scalable unbiased sampling method based on multi-peer adaptive random walk

引用

Ruan Jian Xue Bao/Journal of Software 2009年第3期20卷 630-643页

作者： Fu, Yong-Quan Wang, Yi-Jie Zhou, Jing National Key Laboratory for Parallel and Distributed Processing School of Computer National University of Defense Technology Changsha 410073 China Staff Room of Computer Command Academy of the Corps of Engineers Xuzhou 221000 China

To deal with the scalable and fast unbiased sampling problems in unstructured P2P systems, a sampling method based on multi-peer adaptive random walk (SMARW) is proposed. In the method, based on the multi-peer random walk process, a set of provisional peers are selected as agents which start the sampling processes, by which the sampling process is speeded up with receiving a set of tunable number samples each time;Meanwhile, after receiving new samples earlier agents are replaced with these new samples which repeat the sampling process. With this simple replacement, it can be guaranteed with high probability that the system can reach the optimal load balance;furthermore, SMARW adopts an adaptive distributed random walk adjustment process to increase the convergence rate of the sampling process. A detailed theorical analysis and performance evaluation confirm that SMARW has a high level of unbiased sampling and near-optimal load balancing capability. © by Institute of Software, the Chinese Academy of Sciences. All rights reserved.

关键词： Peer to peer networks

来源：评论

学校读者我要写书评

暂无评论

RLLL: Accurate Relative Localization of RFID Tags with Low Latency

RLLL: Accurate Relative Localization of RFID Tags with Low L...

引用

International Workshop on Quality of Service

作者： Xuan Liu Quan Yang Shigeng Zhang Bin Xiao The Science and Technology on Parallel and Distributed Processing Laboratory (PDL) P. R. China College of Computer Science and Electronic Engineering Hunan University P. R. China School of Computer Science and Engineering Central South University P. R. China The Hong Kong Polytechnic University Hong Kong

ISBN: (数字)9781728168876

ISBN: (纸本)9781728168883

Radio frequency identification (RFID) has been widely used in many smart applications. In many scenarios, it is essential to know the ordering of a set of RFID tags. For example, to quickly detect misplaced books in smart libraries, we need to know the relative ordering of the tags attached to the books. Although several relative RFID localization algorithms have been proposed, they usually suffer from large localization latency and cannot support applications that require real-time detection of tag (product) positions like automatic manufacturing on an assembly line. Moreover, existing approaches face significant degradation in ordering accuracy when the tags are close to each other. In this paper, we propose RLLL, an accurate Relative Localization algorithm for RFID tags with Low Latency. RLLL reduces localization latency by proposing a novel geometry-based approach to identifying the V-zone in the phase reading sequence of each tag. Moreover, RLLL uses only the data in the V-zone to calculate relative positions of tags and thus avoids the negative effects of low-quality data collected when the tag is far from the antenna. Experimental results with commercial RFID devices show that RLLL achieves an ordering accuracy of higher than 0.986 with latency less than 0.8 seconds even when the tags are spaced only 7 mm from adjacent tags, in which case the state-of-the-art solutions only achieve ordering accuracy of lower than 0.8 with localization latency larger than 3 seconds.

关键词： RFID tags Manufacturing Antenna arrays Quality of service Libraries

来源：评论

学校读者我要写书评

暂无评论

On Implementation of a scalable wallet-size cluster computing system for multimedia applications

Lecture Notes in Computer Science (including subseries Lectu...

引用

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 2004年 3333卷 697-704页

作者： Lee, Liang-Teh Li, Kuan-Ching Yang, Chao-Tung Tseng, Chia-Ying Liu, Kang-Yuan Hung, Chih- Dept. of Computer Science and Engineering Tatung University Taipei 104 Taiwan Parallel and Distributed Processing Center Dept. of Computer Science and Information Management Providence University Taichung 433 Taiwan High Performance Computing Laboratory Dept. of Computer Science and Information Engineering Tunghai University Taichung 407 Taiwan Networks and Multimedia Institute Institute for Information Industry Taipei 106 Taiwan

In order to obtain a low-cost, low-power consumption and small sized digital home processing center, especially for multimedia applications, we propose a scalable wallet-size cluster computing system in this paper. Such system uses low-cost, low-power consumption and small sized microprocessors as computing nodes to form a cluster system. The main reason why we are interested to investigate and build such system is to increase the computing power and to accelerate the H.264 video processing. We analyzed the H.264 decoder scheme and distributed the most heavy-load computations in the decoder scheme to each computing node (or microprocessor) of the cluster system for the processing of signals, in parallel. To run multimedia applications in this cluster system, an efficient real-time video processing is implemented. © Springer-Verlag 2004.

关键词： Computing power

来源：评论

学校读者我要写书评

暂无评论

Symmetric Non-negative Matrix Factorization Based Link Partition Method for Overlapping Community Detection

Symmetric Non-negative Matrix Factorization Based Link Parti...

引用

IEEE International Conference on Systems, Man and Cybernetics

作者： Xiang Zhang Naiyang Guan Wenju Zhang Xuhui Huang Shuyi Wu Zhigang Luo Science and Technology on Parallel and Distributed Processing Laboratory College of Computer National University of Defense Technology Hunan P.R. China Institute of Software College of Computer National University of Defense Technology Hunan P.R. China Department of Computer Science and Technology College of Computer National University of Defense Technology Hunan P.R. China

ISBN: (纸本)9781479986989

Partitioning links rather than nodes is effective in overlapping community detection (OCD) on complex networks. However, it consumes high CPU and memory overheads because the volume of links is huge especially when the network is rather complex. In this paper, we proposes a symmetric non-negative matrix factorization (SNMF) based link partition method called SNMF-Link to overcome this deficiency. In particular, SNMF-Link represents data in a lower-dimensional space spanned by the node-link incidence matrix. By solving a lighter SNMF problem, SNMF-Link learns the clustering indicators of each links. Since traditional multiplicative update rule (MUR) based optimization algorithm for SNMF suffers from slow convergence, we applied the augmented Lagrangian method (ALM) to efficiently optimize SNMF. Experimental results show that SNMF-Link is much more efficient than the representative clustering algorithms without reducing the OCD performance.

关键词： Symmetric matrices Convergence Optimization Partitioning algorithms Chlorine Complex networks Image edge detection

来源：评论

学校读者我要写书评

暂无评论

引用

International Conference on parallel and distributed Systems (ICPADS)

作者： Saiqin Long Zhetao Li Zihao Liu Qingyong Deng Sangyoon Oh Nobuyoshi Komuro Key Laboratory of Hunan Province for Internet of Things and Information Security Hunan International Scientific and Technological Cooperation Base of Intelligent network School of Computer Science Xiangtan University Xiangtan China Science and Technology on Parallel and Distributed Processing Laboratory (PDL) National University of Defense Technology Changsha China Ajou University Suwon South Korea Graduate School of Advanced Integration Science Chiba University Chiba Japan

ISBN: (数字)9781728190747

ISBN: (纸本)9781728183824

Deduplication is a data redundancy elimination technique, designed to save system storage resources by reducing redundant data in cloud storage systems. With the development of cloud computing technology, deduplication has been increasingly applied to cloud data centers. However, traditional technologies face great challenges in big data deduplication to properly weigh the two conflicting goals of deduplication throughput and high duplicate elimination ratio. This paper proposes a similarity clustering-based deduplication strategy (named SCDS), which aims to delete more duplicate data without significantly increasing system overhead. The main idea of SCDS is to narrow the query range of fingerprint index by data partitioning and similarity clustering algorithms. In the data preprocessing stage, SCDS uses data partitioning algorithm to classify similar data together. In the data deletion stage, the similarity clustering algorithm is used to divide the similar data fingerprint superblock into the same cluster. Repetitive fingerprints are detected in the same cluster to speed up the retrieval of duplicate fingerprints. Experiments show that the deduplication ratio of SCDS is better than some existing similarity deduplication algorithms, but the overhead is only slightly higher than some high throughput but low deduplication ratio methods.

关键词： Cloud computing Clustering algorithms Estimation Fingerprint recognition Throughput Partitioning algorithms Classification algorithms

来源：评论

学校读者我要写书评

暂无评论

Two layered P2P model for semantic service discovery

引用

Ruan Jian Xue Bao/Journal of Software 2007年第8期18卷 1922-1932页

作者： Liu, Zhi-Zhong Wang, Huai-Min Zhou, Bin Institute of Network and Information Security School of Computer National University of Defense Technology Changsha 410073 China National Laboratory for Parallel and Distributed Processing School of Computer National University of Defense Technology Changsha 410073 China

In open Internet environment, it is inevitable that multiple ontologies coexist. Centralized service discovery mechanism becomes the bottleneck of SOC (service oriented computing), which results in poor scalability of system. Aiming at solving these problems, a two layered P2P based model for semantic service discovery is proposed in this paper. The model is based on ontology community and integrates iVCE (Internet-based virtual computing environment) core concepts into a P2P model. Based on this model, a service discovery algorithm composed of two stages and three steps is proposed. It matches services across communities as well as within community. Within a community, algorithm locates registers holding service information with a high probability of satisfying a request firstly. Then it captures semantic matching between service advertisements and service requests by logical reasoning. Service discovery across communities occurs according to some policies. The model is suitable for opening environment with coexistent multiple ontologies. Experimental results show that given an appropriate setting, the model can make a tradeoff between recall and responding time. In addition, the model will release the mean load of registers efficiently while holding recall.

关键词： Semantics

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：