检索结果-内蒙古大学图书馆

International Symposium on Software Reliability engineering (ISSRE)

作者： Mingyue Zhang Zhi Jin Jian Hou Renwei Luo Schhool of Computer Science Peking University Beijing China Key Lab of High Confidence Software Technologies (PKU) MoE China School of Information Science & Technology Zhejiang Sci-Tech University Hangzhou China School of Computer Science and Engineering University of Electronic Science and Technology Chengdu China

ISBN: (数字)9781665451321

ISBN: (纸本)9781665451338

Distributed deep reinforcement learning(DDRL) has been used in distributed systems to better improve the adaptability. However, DDRL-based systems are also inevitably under the threat of Byzantine workers. There is an urgent need to enhance the resilience of the DDRL-based system against Byzantine failures. This paper proposes a resilient mechanism for mitigating the influence of Byzantine workers on DDRL-based systems. First, we formalize the DDRL-based system as a multi-armed bandit model for well capturing the collective effect of workers on the whole learning process, and then transforming the resilient mechanism design problem into the sampling policy optimization problem. Second, we propose a self-adaptation process for filtering out the harmful data generated by Byzantine workers and theoretically give a mathematical analysis of the understanding, demonstrating its effectiveness under ideal conditions. Third, based on a typical DDRL-based system (i.e., Asynchronous Advantage Actor-Critic, A3C), we implement a resilient distributed A3C (ReD-A3C). With extensive experiments on the DDRL benchmark tasks, we show that ReD-A3C outperforms available Byzantine tolerant approaches.

关键词： Deep learning Filtering Mathematical analysis Reinforcement learning Benchmark testing Robustness Software reliability

来源：评论

学校读者我要写书评

暂无评论

A Hybrid scheme for Parsing Cantonese Text Based on PyCantonese Plus and PyLTP

A Hybrid scheme for Parsing Cantonese Text Based on PyCanton...

引用

Natural Language Processing and Information Retrieval (ECNLPIR), European Conference on

作者： Chunxiao Huang Chunyu Li Shaowen Yao Ye Ding Mingsuo Bao Kun She School of Chinese Language and Literature Yunnan University Kunming China School of Math&Information Science Guangzhou University Guangzhou China National Pilot School of software Yunnan University Kunming China School of Cyberspace Security Dongguan University of Technology Dongguan China College of Computer Sci&Engineering Univ. of Electronic Sci&Tech of China Chengdu China

ISBN: (纸本)9781665473835

A novel scheme for the Cantonese text parsing based on the method of dynamically expanding corpus is presented. Scheme need prepare an additional Chat format file, i.e., export a Cantonese phase file from Cantonese Dictionary firstly, which is widely used in Chinese Mainland, then annotate Jyutping romanization online and annotate them with part-of-speech (commonly known as POS) tags, finally generate the additional Chat format file based on CHILDES database and the above-mentioned Cantonese phase file Jyutping file, POS file, etc. When parsing Cantonese text, we execute iterative operations such as word segmentation with PyCantonese, expanding corpus until the segment result is ideal. Then we do other tasks including POS tagging with our POS dictionary and PyCantonese‘s pos-tagging module, dependency parsing with PyLTP, and visual display all the above parsing results, etc. Test results show the superior performance of the scheme and potential for the parsing Cantonese text.

关键词： Visualization Dictionaries Databases Europe Tagging Natural language processing Iterative methods

来源：评论

学校读者我要写书评

暂无评论

Dart: Domain-adversarial residual-transfer networks for unsupervised cross-domain image classification

arXiv

引用

arXiv 2018年

作者： Fang, Xianghong Bai, Haoli Guo, Ziyi Shen, Bin Hoi, Steven Xu, Zenglin SMILE Lab Univ. of Electronic Sci and Tech.of China Chengdu Sichuan611731 China Dept Computer Science and Engineering Chinese University of Hong Kong Shatin N. T. Hong Kong SMILE Lab Sch. of Computer Science and Engineering University of Electronic Science and Technology of China Chengdu Sichuan611731 China Google Inc. Singapore Management University

The accuracy of deep learning (e.g., convolutional neural networks) for an image classification task critically relies on the amount of labeled training data. Aiming to solve an image classification task on a new domain that lacks labeled data but gains access to cheaply available unlabeled data, unsupervised domain adaptation is a promising technique to boost the performance without incurring extra labeling cost, by assuming images from different domains share some invariant characteristics. In this paper, we propose a new unsupervised domain adaptation method named Domain-Adversarial Residual-Transfer (DART) learning of Deep Neural Networks to tackle cross-domain image classification tasks. In contrast to the existing unsupervised domain adaption approaches, the proposed DART not only learns domain-invariant features via adversarial training, but also achieves robust domainadaptive classification via a residual-transfer strategy, all in an end-to-end training framework. We evaluate the performance of the proposed method for cross-domain image classification tasks on several well-known benchmark data sets, in which our method clearly outperforms the state-of-the-art approaches. Copyright © 2018, The Authors. All rights reserved.

关键词： Image classification

来源：评论

学校读者我要写书评

暂无评论

R-Vivaldi: Distance range-aware IP network coordinate system

引用

Tongxin Xuebao/Journal on Communications 2012年第2期33卷 87-92页

作者： Wang, Da-Bin Huang, Qiong Yang, Xiao-Long Long, Ke-Ping Key Lab. of Mobile Comm. Tech. Chongqing University of Posts and Telecommunications Chongqing 400065 China Research Center for Optical Internet and Mobile Info. Netw. Univ. of Electronic Sci. and Tech. of China Chengdu 611731 China School of Computer and Communications Engineering University of Science and Technology Beijing Beijing 100083 China

The experience showed that the predicted delay's accuracy was improved when the nodes were chosen around the given delay as anchors. Thus a distance range-aware IP network coordinate system(R-Vivaldi) was proposed. The main idea was: according to the approximate range of the predicted delay, the anchors were re-selected in the radius area of the approximate range to compute the new range. In accordance with the above process, the coordinate system constantly iterated the network distance and dynamically chose anchors until the prediction accuracy meet the requirement. Experiments show that the approach can not only effectively improve the prediction accuracy of delay but also overcome inconsistencies of the prediction accuracy to a certain extent.

关键词： Forecasting

来源：评论

学校读者我要写书评

暂无评论

Influence of stochastic delay contaminations on IP network coordinate system and contamination mitigation methods

引用

Tongxin Xuebao/Journal on Communications 2011年第12期32卷 29-35页

作者： Zhang, Min Zhou, Liang Huang, Qiong Yang, Xiao-Long Long, Ke-Ping School of Computer and Comm. Engineering Beijing University of Science and Technology Beijing 100083 China School of Comm. and Info. Engineering Univ. of Electronic Sci. and Tech. of China Chengdu 611731 China School of Comm. and Info. Engineering Chongqing Univ. of Posts and Telecomm. Chongqing 400065 China

There are various stochastic delay contaminations in the IP networks, which have a great impact on the performance of IP network coordinates system in terms of the delay prediction accuracy and the coordinate stability. Here, from the perspective of keeping the statistical characteristic of delay without a loss and suppressing the abnormal delay jitter, two contamination mitigation methods were proposed respectively, i. e., t-Filter and Δ-Filter. Through the PlanetLab datasets, compared with MP-Filter, the simulation results show that both t-Filter and Δ-Filter can enhance the delay prediction accuracy, the system stability and the system convergence in the more effective methods.

关键词： System stability

来源：评论

学校读者我要写书评

暂无评论

ISM: Intent-perceptible service migration model for IP network survivability

引用

Tien Tzu Hsueh Pao/Acta electronica Sinica 2011年第12期39卷 2768-2775页

作者： Zhao, Er-Hu Yang, Xiao-Long Xu, Jie Long, Ke-Ping Zhang, Dan Research Centre for Optical Internet and Mobile Information Networks University of Electronic Science and Technology of China Chengdu Sichuan 611731 China National Key Lab. of Integrated Information System Tech. Institute of Software Chinese Acad. of Sci. Beijing 100190 China School of Computer and Communication Engineering University of Science and Technology Beijing Beijing 100083 China 706 Institute The Second Academy of China Aerospace Beijing 100854 China

The existing service migration methods bring the server cluster heavy burden, and their time spending is so high that seriously affects the service continuity. In order to enhance the survivability of IP network services, the paper puts forward an intent-perceptible service migration model (ISM). In this model, we design a novel trigger mechanism and object selecting method for service migration, and make the service migration intention perceptible for the client with the cooperation of the server. The evaluation results show that ISM model can not only ensure the service invulnerability by maintaining a high randomness of service migration, but also improve the service continuity by reducing the service gap time spending on service migration to the maximum extent. Therefore, ISM model can enhance IP network survivability efficiently.

关键词： Internet protocols

来源：评论

学校读者我要写书评

暂无评论

Rate control scheme for H.264/AVC video encoding

Rate control scheme for H.264/AVC video encoding

引用

2010 International Conference on computer Application and System Modeling, ICCASM 2010

作者： Zheng, Lihua Wu, Yue Tian, Ling Wu, Shuang School of Computer Science and Engineering University of Electronic Sci. and Tech. of China Chengdu 610054 China College of Computer Science Zhejiang University Hangzhou 310058 China

ISBN: (纸本)9781424472369

Rate control is a critical part of video compression systems, while the introduction of Rate-Distortion Optimization makes rate control in H.264/AVC more complex than previous standards. The frame coding complexity MAD is predicted in H.264/AVC rate control. In this paper, we present a novel estimation method for frame coding complexity in H.264/AVC. Meanwhile, a rate control scheme is proposed to improve the coding efficiency. Experimental results demonstrate that our rate control scheme gains better visual quality than that of joint model in H.264/AVC reference software. © 2010 IEEE.

关键词： Image coding

来源：评论

学校读者我要写书评

暂无评论

RATE CONTROL SCHEME FOR H.264/AVC VIDEO ENCODING

RATE CONTROL SCHEME FOR H.264/AVC VIDEO ENCODING

引用

The 2010 International Conference on computer Application and System Modeling(2010计算机应用与系统建模国际会议 ICCASM 2010)

作者： Lihua Zheng Ling Tian Yue Wu Shuang Wu School of Computer Science & Engineering University of Electronic Sci.& Tech.Of China Chengdu P.R.C College of Computer Science Zhejiang University Hangzhou P.R.China 310058

ISBN: (纸本)9781424472352

关键词： R-D model video coding rate control H.264/A VC

来源：评论

学校读者我要写书评

暂无评论

Chaotic and pattern recognition properties of a network of Logistic Neurons

Chaotic and pattern recognition properties of a network of L...

引用

2010 2nd International Conference on computer engineering and technology, ICCET 2010

作者： Ke, Qin Oommen, B. John Lab of Computer Reasoning and Trust-worthy Computation School of Computer Science and Engineering University of Electronic Sci and Tech of China Chengdu Sichuan 610054 China IAPR School of Computer Science Carleton University Ottawa ON K1S5B6 Canada

ISBN: (纸本)9781424463503

The literature on chaos theory reports numerous Neural Networks (NNs) in which the individual neurons have been interconnected and designed so that the overall network yields chaotic behaviour. Although such Chaotic Neuron Net-works(CNNs) have been extensively studied, the results concerning CNNs which can demonstrate chaos, Associative Memory (AM), and Pattern Recognition (PR) are scanty. Recently, the Adachi NN and its variants have been shown to yield the entire spectrum of these properties as its/their parameters change. In this vein, we1 investigate the properties of a specific network of Logistic Neurons (LNs). By appropriately defining the input/output characteristics of a fully connected network of LNs, and by defining their set of weights and output functions, we have succeeded in designing a Logistic Neural Network (LNN) possessing some of these properties. Indeed, by varying the parameters of the LNN, we show that it can yield AM and PR properties for different settings of the parameters. As far as we know, the results presented here novel, and represent the first NN (other than the Adachi NN) which simultaneously exhibits both chaotic, AM and PR properties. © 2010 IEEE.

关键词： Neural networks

来源：评论

学校读者我要写书评

暂无评论

无线传感器网络研究与发展

无线传感器网络研究与发展

引用

电子科技大学电子科学技术研究院第四届学术会议

作者： Ke Zhang 张可 Wei Zhang 张伟 Jiazhi Zeng 曾家智 Research Institute of Electronic Science and Technology of Univ.of Electron.Sci. & Tech.of China Che 电子科技大学电子科学技术研究院成都 610054 电子科技大学计算机科学与工程学院成都 610054 Research Institute of Electronic Science and Technology of Univ.of Electron.Sci. & Tech.of China Che 电子科技大学电子科学技术研究院成都 610054 School of Computer Science and Engineering of Univ.of Electron. Sci.&Tech. of China Chengdu 610054 电子科技大学计算机科学与工程学院成都 610054

本文简要介绍了无线传感器网络的军事领域应用、网络结构、节点组成，同时分析了无线传感器网络的特点及其与现有网络的区别。进而介绍现有无线传感器网络中的MAC层技术、路由技术、节点技术和数据融合等关键技术。最后展望无线传感器... 详细信息

本文简要介绍了无线传感器网络的军事领域应用、网络结构、节点组成，同时分析了无线传感器网络的特点及其与现有网络的区别。进而介绍现有无线传感器网络中的MAC层技术、路由技术、节点技术和数据融合等关键技术。最后展望无线传感器网络的应用和发展并指出关键技术的进步将起到的促进作用。

关键词：无线传感器网络军事应用网络结构节点设计特性分析

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：