检索结果-内蒙古大学图书馆

15th International Conference on Information Security Practice and Experience (ISPEC)

作者： Ciucanu, Radu Lafourcade, Pascal Lombard-Platet, Marius Soare, Marta Univ Orleans LIFO EA 4022 INSA Ctr Val de Loire Orleans France Univ Clermont Auvergne LIMOS CNRS UMR 6158 Aubiere France PSL Res Univ CNRS Ecole Normale Super Dept Informat Paris France Be Studys Geneva Switzerland

ISBN: (纸本)9783030343392;9783030343385

The stochastic multi-armed bandit is a classical decision making model, where an agent repeatedly chooses an action (pull a bandit arm) and the environment responds with a stochastic outcome (reward) coming from an unknown distribution associated with the chosen action. A popular objective for the agent is that of identifying the arm with the maximum expected reward, also known as the best-arm identification problem. We address the inherent privacy concerns that occur in a best-arm identification problem when outsourcing the data and computations to a honest-but-curious cloud. Our main contribution is a distributed protocol that computes the best arm while guaranteeing that (i) no cloud node can learn at the same time information about the rewards and about the arms ranking, and (ii) by analyzing the messages communicated between the different cloud nodes, no information can be learned about the rewards or about the ranking. In other words, the two properties ensure that the protocol has no security single point of failure. We rely on the partially homomorphic property of the well-known Paillier's cryptosystem as a building block in our protocol. We prove the correctness of our protocol and we present proof-of-concept experiments suggesting its practical feasibility.

关键词： Multi-armed bandits Best arm identification Privacy distributed computation Paillier cryptosystem

来源：评论

学校读者我要写书评

暂无评论

LT Codes with Feedback: Accelerate the distributed Matrix-Vector Multiplication with Stragglers 38

LT Codes with Feedback: Accelerate the Distributed Matrix-Ve...

引用

IEEE 38th International Performance Computing and Communications Conference (IPCCC)

作者： Yang, Xiao Jiang, Ming Zhao, Chunming Southeast Univ Natl Mobile Commun Res Lab Nanjing Peoples R China Purple Mt Lab Nanjing Peoples R China

ISBN: (纸本)9781728110257

In this paper, we propose a coding scheme for distributed matrix-vector multiplication that builds upon the Luby transform (LT) codes with feedback. The ideal soliton distribution is utilized in our LT coding scheme to encode the sub-matrices. Besides, the belief propagation (BP) decoding algorithm is modified to cooperate with the feedback information. Compared with other coded distributed computations with straggling servers, our approach achieves lower computation latency when the overall delay incurred by the encoding, mapping and decoding process is considered. Furthermore, we compare the storage loads of different schemes and show that the LT coding with feedback has a strong comparative advantage in these straggler-tolerant computation scenarios.

关键词： rateless feedback codes distributed computation stragglers

来源：评论

学校读者我要写书评

暂无评论

Content Compression Coding for Federated Learning 11

Content Compression Coding for Federated Learning

引用

11th IEEE International Conference on Wireless Communications and Signal Processing (WCSP)

作者： Deng, Kaihe Chen, Zhikun Zhang, Sihai Gong, Chen Zhu, Jinkang Univ Sci & Technol China Key Lab Wireless Opt Commun Hefei Anhui Peoples R China Univ Sci & Technol China PCNSS Hefei Anhui Peoples R China

ISBN: (纸本)9781728135557

As machine learning requires distributed framework, Federated Learning(FL) is proposed to address the distributed computations in machine learning, especially for privacy protection. In FL, an important problem is to reduce the transmission quantity between the terminals and the centralized server. In this paper the content compression coding for FL is proposed, which can effciently reduce the required transmission bandwidth. The correlation among the parameters of the distributed ANN models is first verified based on the entropy analysis and simulation results validate the propsed content compression coding using the WinRAR compression. Our work demonstrates that the proposed coding can compress the parameters in the distributed ANN with acceptable learning performance.

关键词： Federated Learning Data Compression distributed computation

来源：评论

学校读者我要写书评

暂无评论

Building Large-Scale Deep Learning System for Entity Recognition in E-Commerce Search 19

Building Large-Scale Deep Learning System for Entity Recogni...

引用

12th IEEE/ACM Conference on Utility and Cloud Computing (UCC) / 6th IEEE/ACM International Conference on Big Data Computing, Applications and Technologies (BDCAT)

作者： Wen, Musen Vasthimal, Deepak Kumar Lu, Alan Wang, Tian Guo, Aimin eBay Inc San Jose CA 95125 USA

ISBN: (纸本)9781450370165

Named-Entity-Recognition (NER) or Item Aspect Recognition task is fundamental to e-commerce marketplace. A structured listing (i.e. item aspect and catalog) is critical to the success of the marketplace - it helps the seller to put their listings to the right catalog and aspects so that the buyers can easily find the most relevant and accurate listings they are looking for. For e-commerce search engine, item aspects (brand, color, size, texture, etc.) should be automatically recognized from users' shopping queries in order to accurate understand their shopping intent. An accurate query and item aspect understanding helps to match seller's listing to buyer's purchase intend. However, in practice, this still remains challenge in e-commerce marketplace due to a couple reasons, e.g. the sparsity of the data-hundreds of millions of item aspect-name and aspect-value pairs (e.g. brand=Nike);noisy (low quality), not scalable but expensive label data that are obtained via human labeling [12] effort;the lack of context in generally very short search queries;the disperse of context and dis-orderedness in listing title or descriptions, etc. among others. All those imposes a big challenge that affect the performance of aspect recognition for e-commerce practice. In this paper, we introduce an end-to-end machine learning system to build an effective aspect recognition system for search engine that leverages the generated label data from different legacy systems and effort, without any extra human-label effort (and thus no incurred cost). The framework is constructed with multiple machine learning components that are built to optimize the search relevance and conversion as an end goal. We show that the proposed aspect recognition machine learning system improved search results quality and relevance greatly compared to the existing already strong baseline system for e-commerce search.

关键词： name-entity-recognition aspect recognition e-commerce neural networks text mining deep learning big data artificial intelligence parallel processing distributed computation

来源：评论

学校读者我要写书评

暂无评论

The Random Component-Wise Power Method 18

The Random Component-Wise Power Method

引用

Conference on Wavelets and Sparsity XVIII

作者： Teke, Oguzhan Vaidyanathan, Palghat P. CALTECH 1200 E Calif Blvd Pasadena CA 91125 USA

ISBN: (数字)9781510629707

ISBN: (纸本)9781510629707

This paper considers a random component-wise variant of the unnormalized power method, which is similar to the regular power iteration except that only a random subset of indices is updated in each iteration. For the case of normal matrices, it was previously shown that random component-wise updates converge in the mean-squared sense to an eigenvector of eigenvalue 1 of the underlying matrix even in the case of the matrix having spectral radius larger than unity. In addition to the enlarged convergence regions, this study shows that the eigenvalue gap does not directly affect the convergence rate of the randomized updates unlike the regular power method. In particular, it is shown that the rate of convergence is affected by the phase of the eigenvalues in the case of random component-wise updates, and the randomized updates favor negative eigenvalues over positive ones. As an application, this study considers a reformulation of the component-wise updates revealing a randomized algorithm that is proven to converge to the dominant left and right singular vectors of a normalized data matrix. The algorithm is also extended to handle large-scale distributed data when computing an arbitrary rank approximation of an arbitrary data matrix. Numerical simulations verify the convergence of the proposed algorithms under different parameter settings.

关键词： Fixed point iteration randomized iterations low-rank approximation distributed computation

来源：评论

学校读者我要写书评

暂无评论

Feature Selection with Evolving, Fast and Slow Using Two Parallel Genetic Algorithms 4

Feature Selection with Evolving, Fast and Slow Using Two Par...

引用

4th International Conference on Computer Science and Engineering (UBMK)

作者： Cetin, Uzay Gundogmus, Yunus Emre Istanbul Bilgi Univ Comp Engn Dept Istanbul Turkey Istanbul Marmara Univ Stat Dept Tam Faktoring Istanbul Turkey

ISBN: (纸本)9781728139647

Feature selection is one of the most challenging issues in machine learning, especially while working with high dimensional data. In this paper, we address the problem of feature selection and propose a new approach called Evolving Fast and Slow. This new approach is based on using two parallel genetic algorithms having high and low mutation rates, respectively. Evolving Fast and Slow requires a new parallel architecture combining an automatic system that evolves fast and an effortful system that evolves slow. With this architecture, exploration and exploitation can be done simultaneously and in unison. Evolving last, with high mutation rate, can be useful to explore new unknown places in the search space with long jumps;and Evolving Slow, with low mutation rate, can be useful to exploit previously known places in the search space with short movements. Our experiments show that Evolving Fast and Slow achieves very good results in terms of both accuracy and feature elimination.

关键词： Feature selection Genetic Algorithms High Di-mensional Data distributed computation Machine Learning

来源：评论

学校读者我要写书评

暂无评论

Model Inversion Attacks Against Collaborative Inference 19

Model Inversion Attacks Against Collaborative Inference

引用

35th Annual Computer Security Applications Conference (ACSA)

作者： He, Zecheng Zhang, Tianwei Lee, Ruby B. Princeton Univ Princeton NJ 08544 USA Nanyang Technol Univ Singapore Singapore

ISBN: (纸本)9781450376280

The prevalence of deep learning has drawn attention to the privacy protection of sensitive data. Various privacy threats have been presented, where an adversary can steal model owners' private data. Meanwhile, countermeasures have also been introduced to achieve privacy-preserving deep learning. However, most studies only focused on data privacy during training, and ignored privacy during inference. In this paper, we devise a new set of attacks to compromise the inference data privacy in collaborative deep learning systems. Specifically, when a deep neural network and the corresponding inference task are split and distributed to different participants, one malicious participant can accurately recover an arbitrary input fed into this system, even if he has no access to other participants' data or computations, or to prediction APIs to query this system. We evaluate our attacks under different settings, models and datasets, to show their effectiveness and generalization. We also study the characteristics of deep learning models that make them susceptible to such inference privacy threats. This provides insights and guidelines to develop more privacy-preserving collaborative systems and algorithms.

关键词： Deep Neural Network Model Inversion Attack distributed computation

来源：评论

学校读者我要写书评

暂无评论

distributed Matrix-Vector Multiplication: A Convolutional Coding Approach

Distributed Matrix-Vector Multiplication: A Convolutional Co...

引用

IEEE International Symposium on Information Theory (ISIT)

作者： Das, Anindya B. Ramamoorthy, Aditya Iowa State Univ Dept Elect & Comp Engn Ames IA 50010 USA

ISBN: (纸本)9781538692912

distributed computing systems are well-known to suffer from the problem of slow or failed nodes;these are referred to as stragglers. Straggler mitigation (for distributed matrix computations) has recently been investigated from the standpoint of erasure coding in several works. In this work we present a strategy for distributed matrix-vector multiplication based on convolutional coding. Our scheme can be decoded using a low-complexity peeling decoder. The recovery process enjoys excellent numerical stability as compared to Reed-Solomon coding based approaches (which exhibit significant problems owing their badly conditioned decoding matrices). Finally, our schemes are better matched to the practically important case of sparse matrix-vector multiplication as compared to many previous schemes. Extensive simulation results corroborate our findings.

关键词： distributed computation Stragglers Cross Parity Check Convolutional Code Reed-Solomon Coding

来源：评论

学校读者我要写书评

暂无评论

A distributed Fuzzy Support Vector Machines Model for Real Network Traffic 9

A Distributed Fuzzy Support Vector Machines Model for Real N...

引用

2019 9th International Workshop on Computer Science and Engineering, WCSE 2019

作者： JIANG Jie QU Hua ZHAO Jihong ZHANG Yanpeng Xi'an Jiaotong University Xi'an University of Posts & Telecommunications

ISBN: (纸本)9789811416842

Traffic classification has been widely applied for *** works paid little attention to robustness and massive data of practice network *** this paper,we propose a new strategy for the Random Fuzzy Support Vector Machines(RA-FSVM) based on fuzzy membership,as well as employ genetic algorithm to find optimal ***,we introduce the distributed idea to structure clusters,so as to speed up RA-FSVM model's *** results of simulation demonstrate the robustness and timeliness of RA-FSVM model.

关键词： Network traffic Fuzzy Support Vector Machines Genetic algorithm distributed computation

来源：评论

学校读者我要写书评

暂无评论

A Novel Bitcoin Mining Scheme Based on the Multi-Leader Multi-Follower Stackelberg Game Model

引用

IEEE ACCESS 2018年 6卷 48902-48912页

作者： Kim, Sungwook Sogang Univ Dept Comp Sci Seoul 121742 South Korea

Since the introduction of Bitcoin in 2009, it has gained a significant popularity around the world. Bitcoin is a peer-to-peer electronic payment system where payment transactions are stored in a data structure named the block-chain. Based on a fully decentralized network, the blockchain is maintained by a community of participants. In Bitcoin system, mining is the fundamental concept. In this paper, we design a new Bitcoin mining scheme based on the multi-leader multi-follower Stackelberg game model. To effectively implement the peer-to-peer relationship of Bitcoin system agents, we assume that mining pool operators are leaders and mining participating users are followers in our Stackelberg game. By using the dynamics of feedback-based repeated process, leaders, and followers can be interacting with one another and make their decisions in a way to reach an efficient system solution. Without the influence of any central authorities and organizations, the proposed method is practically applied to a distributed Bitcoin system. Through system level simulations, we show that our game approach outperforms the existing Bitcoin schemes in providing a better fair-efficient system performance.

关键词： Bitcoin mining distributed computation peer-to-peer networks multi-leader multi-follower Stackelberg game bargaining solutions

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：