检索结果-内蒙古大学图书馆

International Joint Conference on Neural Networks

作者： Menglong Lu Dawei Feng Linbo Qiao Dawen Ding Dongsheng Li Science and Technology on Parallel and Distributed Laboratory National University of Defense Technology Changsha China College of Computer National University of Defense Technology Changsha China CMC AS2 South CSC Asiainfo Nanchang China

Bundle method for regularized risk minimization (BMRM) is a variant of Cutting Plane Method (CPM). It performs efficiently in solving a convex minimization problem, which is a core part in a plethora of machine learning applications. Nonetheless, while exposed to the challenge of large-scale learning, the synchronous parallel implementation of BMRM easily encounters the straggler problem due to the diversity among heterogeneous working nodes' capability and unevenness in the inherent data distribution. In this paper, we propose a novel asynchronous distributed BMRM implementation, which employs an asynchronous computing window to fully explore the fast nodes' computational capabilities while reserving the good convergence of the BMRM. Extensive experiments show that the asynchronous BMRM algorithm has significant improvement of performance over its synchronous counterpart, and owns the ability to solve large-scale problems efficiently.

关键词： Estimation Linear programming Machine learning Risk management Convergence Servers Minimization

来源：评论

学校读者我要写书评

暂无评论

Design of an Indoor Exploration and Multi-Objective Navigation System

Design of an Indoor Exploration and Multi-Objective Navigati...

引用

Chinese Control Conference (CCC)

作者： Guoxian Zheng Lei Zhang Huaxi Yulin Zhang Bo Ding State Key Laboratory of Synthetical Automation for Process Industries Northeastern University Shenyang China MIS Universite de Picardie Jules Verne Amiens France National Laboratory for Parallel and Distributed Processing National University of Defense Technology Changsha China

In this paper, we propose an indoor robot autonomous navigation system. The robot firstly explores in an unknown environment, and then navigates autonomously by using the explored map. The robot is equipped a 2D laser scanner as the main sensor. The laser scanner is used for path planning and frontier-based exploration. A 2D global occupancy map is built for path planning, frontier-based exploration and multi-objective autonomous navigation. Laser scans are transmitted into Simultaneous Localization and Mapping (SLAM) process in the exploration phase. In indoor environment, the exploration efficiency is improved by merging a heuristic algorithm. By using multi-threading technology and a 3D perception approach proposed in this paper, the robot equipped with a low-cost RGBD sensor can detect all kinds of obstacles to achieve highly reliable navigation in complicated 3D environment. Meanwhile, we develop a multi-objective navigation application to make human-robot interaction more convenient and satisfy multi-task deployment. Our approaches are demonstrated by experimental results.

关键词： Navigation Three-dimensional displays Robot sensing systems Lasers Path planning Two dimensional displays

来源：评论

学校读者我要写书评

暂无评论

Exploiting contention and congestion aware switch allocation in network-on-chips 17

Exploiting contention and congestion aware switch allocation...

引用

50th ACM Turing Conference - China, ACM TUR-C 2017

作者： Li, Cunlu Dongy, Dezun Liao, Xiangke National Laboratory for Parallel and Distributed Processing Collaborative Innovation Center of High Performance Computing College of Computer National University of Defense Technology Changsha Hunan410073 China

ISBN: (纸本)9781450348737

Network-on-chip system plays an important role to improve the performance of chip multiprocessor systems. As the complexity of the network increases, congestion problem has become the major performance bottleneck and seriously influence the performance of NoCs. Prior works have focused on designing effective routing algorithm based on collecting congestion and contention information to load balance the traffic. However, most prior works do not consider balancing the traffic load during switch allocation. Due to the lack of congestion information in switch allocation stage, switch allocator performs allocation only based on packet requests and thus aggravates the congestion in the ports of switch. In this paper, we propose to add the congestion and contention information into the switch allocation process and design an efficient on-chip switching strategy which utilizes the contention and congestion information to load balance the traffic and achieve efficient switch allocation. To further enhance the performance of our design, we carefully design our switch allocation strategy to balance the trade-off between traffic load balance and the matching efficiency in switch allocation. We evaluate our design under synthetic traffic and trace of parsec benchmarks. Our evaluations show this mechanism achieves optimal latency compared to best previous switch allocation strategies. © 2017 ACM.

关键词： Network-on-chip

来源：评论

学校读者我要写书评

暂无评论

Collaborative deep learning across multiple data centers

arXiv

引用

arXiv 2018年

作者： Xu, Kele Mi, Haibo Feng, Dawei Wang, Huaimin Chen, Chuan Zheng, Zibin Lan, Xu National Key Laboratory of Parallel and Distributed Processing Changsha China College of Computer National University of Defense Technology Changsha China School of Data and Computer Science Sun Yat-Sen University Guangzhou China Queen Mary University of London London United Kingdom

Valuable training data is often owned by independent organizations and located in multiple data centers. Most deep learning approaches require to centralize the multi-datacenter data for performance purpose. In practice, however, it is often infeasible to transfer all data to a centralized data center due to not only bandwidth limitation but also the constraints of privacy regulations. Model averaging is a conventional choice for data parallelized training, but its ineffectiveness is claimed by previous studies as deep neural networks are often non-convex. In this paper, we argue that model averaging can be effective in the decentralized environment by using two strategies, namely, the cyclical learning rate and the increased number of epochs for local model training. With the two strategies, we show that model averaging can provide competitive performance in the decentralized mode compared to the data-centralized one. In a practical environment with multiple data centers, we conduct extensive experiments using state-of-the-art deep network architectures on different types of data. Results demonstrate the effectiveness and robustness of the proposed method. Copyright © 2018, The Authors. All rights reserved.

关键词： Network architecture

来源：评论

学校读者我要写书评

暂无评论

Loss rank mining: A general hard example mining method for real-time Detectors

arXiv

引用

arXiv 2018年

作者： Yu, Hao Zhang, Zhaoning Qin, Zheng Wu, Hao Li, Dongsheng Zhao, Jun Lu, Xicheng Science and Technology on Parallel and Distributed Laboratory National University of Defense Technology Changsha China College of Electronic and Engineering National University of Defense Technology Changsha China College of Meteorology and Oceanology National University of Defense Technology Changsha China

Modern object detectors usually suffer from low accuracy issues, as foregrounds always drown in tons of backgrounds and become hard examples during training. Compared with those proposal-based ones, real-time detectors are in far more serious trouble since they renounce the use of regionproposing stage which is used to filter a majority of backgrounds for achieving real-time rates. Though foregrounds as hard examples are in urgent need of being mined from tons of backgrounds, a considerable number of state-of-the-art realtime detectors, like YOLO series, have yet to profit from existing hard example mining methods, as using these methods need detectors fit series of prerequisites. In this paper, we propose a general hard example mining method named Loss Rank Mining (LRM) to fill the gap. LRM is a general method for real-time detectors, as it utilizes the final feature map which exists in all real-time detectors to mine hard examples. By using LRM, some elements representing easy examples in final feature map are filtered and detectors are forced to concentrate on hard examples during training. Extensive experiments validate the effectiveness of our method. With our method, the improvements of YOLOv2 detector on auto-driving related dataset KITTI and more general dataset PASCAL VOC are over 5% and 2% mAP, respectively. In addition, LRM is the first hard example mining strategy which could fit YOLOv2 perfectly and make it better applied in series of real scenarios where both real-time rates and accurate detection are strongly demanded. Copyright © 2018, The Authors. All rights reserved.

关键词： Mining

来源：评论

学校读者我要写书评

暂无评论

Loss Rank Mining: A General Hard Example Mining Method for Real-time Detectors

Loss Rank Mining: A General Hard Example Mining Method for R...

引用

International Joint Conference on Neural Networks

作者： Hao Yu Zhaoning Zhang Zheng Qin Hao Wu Dongsheng Li Jun Zhao Xicheng Lu Science and Technology on Parallel and Distributed Laboratory National University of Defense Technology Changsha China College of Electronic and Engineering National University of Defense Technology Changsha China College of Meteorology and Oceanology National University of Defense Technology Changsha China

Modern object detectors usually suffer from low accuracy issues, as foregrounds always drown in tons of back-grounds and become hard examples during training. Compared with those proposal-based ones, real-time detectors are in far more serious trouble since they renounce the use of region-proposing stage which is used to filter a majority of back-grounds for achieving real-time rates. Though foregrounds as hard examples are in urgent need of being mined from tons of backgrounds, a considerable number of state-of-the-art real-time detectors, like YOLO series, have yet to profit from existing hard example mining methods, as using these methods need detectors fit series of prerequisites. In this paper, we propose a general hard example mining method named Loss Rank Mining (LRM) to fill the gap. LRM is a general method for real-time detectors, as it utilizes the final feature map which exists in all real-time detectors to mine hard examples. By using LRM, some elements representing easy examples in final feature map are filtered and detectors are forced to concentrate on hard examples during training. Extensive experiments validate the effectiveness of our method. With our method, the improvements of YOLOv2 detector on auto-driving related dataset KITTI and more general dataset PASCAL VOC are over 5% and 2% mAP, respectively. In addition, LRM is the first hard example mining strategy which could fit YOLOv2 perfectly and make it better applied in series of real scenarios where both real-time rates and accurate detection are strongly demanded.

关键词： Detectors Real-time systems Feature extraction Training Object detection Pipelines Task analysis

来源：评论

学校读者我要写书评

暂无评论

CloudDPI: Cloud-based privacy-preserving deep packet inspection via reversible sketch 1

引用

9th International Symposium on Cyberspace Safety and Security, CSS 2017

作者： Li, Jie Su, Jinshu Wang, Xiaofeng Sun, Hao Chen, Shuhui School of Computer National University of Defense Technology Changsha Hunan410073 China National Key Laboratory for Parallel and Distributed Processing National University of Defense Technology Changsha Hunan410073 China

ISBN: (数字)9783319694719

ISBN: (纸本)9783319694702

Hardware-based middleboxes are ubiquitous in computer networks, which usually incur high deployment and management expenses. A recently arsing trend aims to address those problems by outsourcing the functions of traditional hardware-based middleboxes to high volume servers in a cloud. This technology is promising but still faces a few challenges. First, the widely adopted data encryption techniques contradict with payload inspection needs of some middleboxes such as DPI and IDS devices. Second, the inspection rules of middleboxes may be commercial properties, thus the middlebox providers want to keep their rules confidential under third-party cloud environments, and this creates hindrances for the cloud to perform outsourced middlebox functions. Third, performance of the outsourced middlebox is an inevitable issue that needs deliberate consideration. In this paper, we propose a cloud-based DPI middlebox implementation which performs payload inspection over encrypted traffic while preserving the privacy of both communication data and inspection rules. Our design employs a modified reversible sketch structure which is used for efficient error-free membership testing, and we utilize unkeyed one-way hash functions instead of complex cryptographic protocols to achieve the privacy preservation requirements. CloudDPI supports a wide range of real-world inspection rules, we conduct evaluations on ClamAV rule set and the experiment results demonstrate the effectiveness of our proposal. © 2017, Springer International Publishing AG.

关键词： Network function virtualization

来源：评论

学校读者我要写书评

暂无评论

Fine-grained checkpoint based on non-volatile memory

引用

Frontiers of Information technology & Electronic Engineering 2017年第2期18卷 220-234页

作者： Wen-zhe ZHANG Kai LU Mikel LUJAN Xiao-ping WANG Xu ZHOU Science and Technology on Parallel and Distributed Processing Laboratory College of Computer National University of Defense Technology Changsha 410072 China School of Computer The University of Manchester Manchester M13 9PL UK

New non-volatile memory （e.g., phase-change memory） provides fast access, large capacity, byteaddressability, and non-volatility features. These features, fast-byte-persistency, will bring new opportunities to fault tolerance. We propose a fine-grained checkpoint based on non-volatile memory. We extend the current virtual memory manager to manage non-volatile memory, and design a persistent heap with support for fast allocation and checkpointing of persistent objects. To achieve a fine-grained checkpoint, we scatter objects across virtual pages and rely on hardware page-protection to monitor the modifications. In our system, two objects in different virtual pages may reside on the same physical page. Modifying one object would not interfere with the other object. This allows us to monitor and checkpoint objects smaller than 4096 bytes in a fine-grained way. Compared with previous page-grained based checkpoint mechanisms, our new checkpoint method can greatly reduce the data copied at checkpoint time and better leverage the limited bandwidth of non-volatile memory.

关键词： Non-volatile memory Byte-persistency Persistent heap Fine-grained checkpoint

来源：评论

学校读者我要写书评

暂无评论

Growing construction and adaptive evolution of complex software systems

引用

science China(Information sciences) 2016年第5期59卷 5-7页

作者： Huaimin WANG Bo DING National Key Laboratory of Parallel and Distributed Processing College of ComputerNational University of Defense Technology

distributed software systems are becoming more and more complex *** is easy to find a huge amount of computing nodes in a nationwide or global information *** example,We Chat(Wei Xin),a well-known mobile application in China,has reached a record of 650 million monthly active users in the third quarter of *** the same time,researchers are starting to talk about software systems which have billions of lines of codes[1]or can last one hundred years.

关键词： Growing construction and adaptive evolution of complex software systems

来源：评论

学校读者我要写书评

暂无评论

parallel Algorithm Design for Audio Feature Extraction

Parallel Algorithm Design for Audio Feature Extraction

引用

2017 5th International Conference on Machinery,Materials and Computing technology(ICMMCT 2017)

作者： Xiaocheng Luo Jingfei Jiang Junjie Zhu Yong Dou National Laboratory for Parallel and Distributed Processing National University of Defense Technology

Audio feature extraction is a very important technique in the field of sound processing. It extremely impacts the effectiveness and correctness of sound recognition, sound verification, etc. It is a computation intensive stage in the whole sound recognition process, which is a challenging for acceleration. In this paper, a coarse-grained parallel feature extraction algorithm for high throughput of audio slices is proposed to improve the efficiency of audio feature extraction. Three typical audio feature extraction algorithms, Mel Frequency Cepstrum Coefficients(MFCC), Spectrogram image features(SIF), Octave-Based Spectral Contrast, are chosen to parallelize. Experiments results on different platforms show that the speedup of accelerated audio feature extraction is up to 17.23 on the platform with 16 cores 32 threads.

关键词： Feature Extraction Sound Recognition parallel Algorithm

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：