检索结果-内蒙古大学图书馆

Efficient Model Store and Reuse in an OLML database System

Journal of Computer Science & Technology 2021年第4期36卷 792-805页

作者： Jian-Wei Cui Wei Lu Xin Zhao Xiao-Yong Du Key Laboratory of Data Engineering and Knowledge Engineering of Ministry of Education Renmin University of China Beijing 100872China School of Information Renmin University of ChinaBeijing 100872China

Deep learning has shown significant improvements on various machine learning tasks by introducing a wide spectrum of neural network ***,for these neural network models,it is necessary to label a tremendous amount of training data,which is prohibitively expensive in *** this paper,we propose OnLine Machine Learning(OLML)database which stores trained models and reuses these models in a new training task to achieve a better training effect with a small amount of training *** efficient model reuse algorithm AdaReuse is developed in the OLML ***,AdaReuse firstly estimates the reuse potential of trained models from domain relatedness and model quality,through which a group of trained models with high reuse potential for the training task could be selected ***,multi selected models will be trained iteratively to encourage diverse models,with which a better training effect could be achieved by *** evaluate AdaReuse on two types of natural language processing(NLP)tasks,and the results show AdaReuse could improve the training effect significantly compared with models training from scratch when the training data is *** on AdaReuse,we implement an OLML database prototype system which could accept a training task as an SQL-like query and automatically generate a training plan by selecting and reusing trained *** studies are conducted to illustrate the OLML database could properly store the trained models,and reuse the trained models efficiently in new training tasks.

关键词： model selection model reuse OnLine Machine Learning(OLML)database

来源：评论

学校读者我要写书评

暂无评论

Efficient protocols for heavy hitter identification with local differential privacy

引用

Frontiers of Computer Science 2022年第5期16卷 193-203页

作者： Dan ZHAO Suyun ZHAO Hong CHEN Ruixuan LIU Cuiping LI Wenjuan LIANG Key Laboratory of Data Engineering and Knowledge Engineering of Ministry of Education Renmin University of ChinaBeijing 100872China School of Information Renmin University of ChinaBeijing 100872China

Local differential privacy(LDP),which is a technique that employs unbiased statistical estimations instead of real data,is usually adopted in data collection,as it can protect every user’s privacy and prevent the leakage of sensitive *** segment pairs method(SPM),multiple-channel method(MCM)and prefix extending method(PEM)are three known LDP protocols for heavy hitter identification as well as the frequency oracle(FO)problem with large ***,the low scalability of these three LDP algorithms often limits their ***,communication and computation strongly affect their ***,excessive grouping or sharing of privacy budgets makes the results *** address the abovementioned problems,this study proposes independent channel(IC)and mixed independent channel(MIC),which are efficient LDP protocols for FO with a large *** design a flexible method for splitting a large domain to reduce the number of ***,we employ the false positive rate with interaction to obtain an accurate *** experiments demonstrate that IC outperforms all the existing solutions under the same privacy guarantee while MIC performs well under a small privacy budget with the lowest communication cost.

关键词： local differential privacy frequency oracle heavy hitter

来源：评论

学校读者我要写书评

暂无评论

Improving Parameter Estimation and Defensive Ability of Latent Dirichlet Allocation Model Training Under Rényi Differential Privacy

引用

Journal of Computer Science & Technology 2022年第6期37卷 1382-1397页

作者： Tao Huang Su-Yun Zhao Hong Chen Yi-Xuan Liu Key Laboratory of Data Engineering and Knowledge Engineering(Renmin University of China) Ministry of Education Beijing 100087China School of Information Renmin University of ChinaBeijing 100087China

Latent Dirichlet allocation(LDA)is a topic model widely used for discovering hidden semantics in massive text *** Gibbs sampling(CGS),as a widely-used algorithm for learning the parameters of LDA,has the risk of privacy ***,word count statistics and updates of latent topics in CGS,which are essential for parameter estimation,could be employed by adversaries to conduct effective membership inference attacks(MIAs).Till now,there are two kinds of methods exploited in CGS to defend against MIAs:adding noise to word count statistics and utilizing inherent *** two kinds of methods have their respective *** sampled from the Laplacian distribution sometimes produces negative word count statistics,which render terrible parameter estimation in *** inherent privacy could only provide weak guaranteed privacy when defending against *** is promising to propose an effective framework to obtain accurate parameter estimations with guaranteed differential *** key issue of obtaining accurate parameter estimations when introducing differential privacy in CGS is making good use of the privacy budget such that a precise noise scale is *** is the first time that R′enyi differential privacy(RDP)has been introduced into CGS and we propose RDP-LDA,an effective framework for analyzing the privacy loss of any differentially private ***-LDA could be used to derive a tighter upper bound of privacy loss than the overestimated results of existing differentially private CGS obtained byε-*** RDP-LDA,we propose a novel truncated-Gaussian mechanism that keeps word count statistics *** we propose distribution perturbation which could provide more rigorous guaranteed privacy than utilizing inherent *** validate that our proposed methods produce more accurate parameter estimation under the JS-divergence metric and obtain lower precision and recall when defending against MIAs.

关键词： latent Dirichlet allocation parameter estimation membership inference attack Rényi differential privacy

来源：评论

学校读者我要写书评

暂无评论

A Timeline-Based Algorithm for Personalized Tag Recommendation

A Timeline-Based Algorithm for Personalized Tag Recommendati...

引用

11th International Conference on Web Information Systems engineering

作者： Yu, Zhaohui Wang, Puwei Du, Xiaoyong Cui, Jianwei Xu, Tianren Key Labs. of Data Engineering and Knowledge Engineering Ministry of Education China School of Information Renmin University of China Beijing 100872 China

ISBN: (纸本)9783642243950

Recently, tagging has been a flexible and important way to share and categorize web resources, these user-generated tags are effective to represent user interests because these tags reflect human being's judgments while more concise and closer to human understanding, and the user interests are changing over time. Thus, modeling user interests to meet individual user needs is an important challenge for personalization and information filtering applications, such as recommender systems. In this paper, we apply a distance decay model for modeling user interests in terms of tags based on timeline. We then propose a novel algorithm to measure users' similarities in terms of their tagging activity over a specific time period and provide personalized tag recommendation according to similar users' interests in their next time intervals. Experimental results demonstrate the higher precision and recall with our personalized tag recommendation algorithm than other existing methods.

关键词： timeline user interests distance decay model personalized tag recommendation

来源：评论

学校读者我要写书评

暂无评论

Study on the Characteristics of Cross-Domain knowledge Diffusion from Science to Policy: Evidence from Overton data

引用

Proceedings of the Association for Information Science and Technology 2023年第1期60卷 368-378页

作者： Ren, Chao Yang, Menghui School of Information Resource Management Renmin University of China China Key Laboratory of Data Engineering and Knowledge Engineering Ministry of Education China

The cross-domain knowledge diffusion from science to policy is a prevalent phenomenon that demands academic attention. To investigate the characteristics of cross-domain knowledge diffusion from science to policy, this study suggests using the citation of policies to scientific articles as a basis for quantifying the diffusion strength, breadth, and speed. The study reveals that the strength and breadth of cross-domain knowledge diffusion from scientific papers to policies conform to a power-law distribution, while the speed follows a logarithmic normal distribution. Moreover, the papers with the highest diffusion strength, breadth, and fastest diffusion speed are predominantly from world-renowned universities, scholars, and top journals. The papers with the highest diffusion strength and breadth are mostly from social sciences, especially economics, those with the fastest diffusion speed are mainly from medical and life sciences, followed by social sciences. The findings indicate that cross-domain knowledge diffusion from science to policy follows the Matthew effect, whereby individuals or institutions with high academic achievements are more likely to achieve successful cross-domain knowledge diffusion. Furthermore, papers in the field of economics tend to have the higher cross-domain knowledge diffusion strength and breadth, while those in medical and life sciences have the faster cross-domain knowledge diffusion speed. 86 Annual Meeting of the Association for Information Science & Technology | Oct. 27 – 31, 2023 | London, United Kingdom. Author(s) retain copyright, but ASIS&T receives an exclusive publication license.

关键词： Diffusion

来源：评论

学校读者我要写书评

暂无评论

Approximation for a scheduling problem with application in wireless networks

引用

Science China Mathematics 2010年第6期53卷 1643-1655页

作者： ZHU QingHua & LI DeYing key Laboratory of data engineering and knowledge engineering,MOE,School of Information,Renmin University of China,Beijing 100872,China 1. Key Laboratory of Data Engineering and Knowledge Engineering MOE School of Information Renmin University of China Beijing 100872 China

A network of many sensors and a base station that are deployed over a region is *** sensor has a transmission range,an interference range and a carrier sensing range,which are r,αr and βr,*** this paper,we study the minimum latency conflict-aware many-to-one data aggregation scheduling problem:Given locations of sensors along with a base station,a subset of all sensors,and parameters r,α and β,to find a schedule in which the data of each sensor in the subset can be transmitted to the base station with no conflicts,such that the latency is *** designe an algorithm based on maximal independent sets,which has a latency bound of(a+19b)R + Δb-a + 5 time slots,where a and b are two constant integers relying on α and β,Δ is the maximum degree of network topology,and R is the trivial lower bound of *** Δ contributes to an additive factor instead of a multiplicative one,thus our algorithm is nearly a constant(a+19b)-ratio.

关键词： approximation algorithm conflict-aware many-to-one data aggregation latency wireless sensor network

来源：评论

学校读者我要写书评

暂无评论

An Adaptive Method for the Efficient Similarity Calculation

An Adaptive Method for the Efficient Similarity Calculation

引用

14th International Conference on database Systems for Advanced Applications

作者： Cai, Yuanzhe Liu, Hongyan He, Jun Du, Xiaoyong Jia, Xu Key Labs of Data Engineering and Knowledge Engineering MOE China School of Information Renmin University of China China Department of Management Science and Engineering Tsinghua University China

ISBN: (纸本)9783642008863

SimRank is a well-known algorithm for similarity calculation based on object-to-object relationship. However, it suffers from high computation cost. Inthis paper, we find that the convergence behavior of different object pairs is different when we use SimRank to compute the similarity of objects. Many similarity scores converge fast, while others need more time before convergence. Based on this observation, we propose an adaptive method called Adaptive-SimRank to speed up similarity calculation. Using this method, we don't need to recalculate those converged pairs' similarity. The experiments conducted on web datasets and synthetic dataset show that our new method can reduce the running time by nearly 35%.

关键词： Similarity Calculation Linkage Mining

来源：评论

学校读者我要写书评

暂无评论

SeCCX: Semantics-based fine granular concurrency control for XML data

SeCCX: Semantics-based fine granular concurrency control for...

引用

1st International Workshop on Graph database, IWGD 2010, 1st International Workshop on Advanced Techniques on XML data Management, XMLDM 2010, 2nd International Workshop on WCMT 2010, Held in Conj. with WAIM 2010

作者： Rong, Chuitian Lu, Wei Zhang, Xiao Liu, Zhen Du, Xiaoyong Key Labs. of Data Engineering and Knowledge Engineering Ministry of Education China School of Information Renmin University of China China

ISBN: (纸本)3642167195

This paper proposes a new locking protocol, SeCCX, for isolation of concurrent transactions on XML data. This protocol adopts the semantics of operations issued by users. Comparing with previous XML locking protocols, SeCCX has some new features, like richer lock modes, finer granular, lower lock confliction and dynamic lock escalation. The importance is that it locks the objects according to the semantics of user's operation and sets different restrictions on different objects. It provides finer granular concurrency control such that the overall system performance can be improved. We prove that the SeCCX protocol is conflict serializability. We also compare it with other protocols to demonstrate the advantages of SeCCX. © 2010 Springer-Verlag.

关键词： Locks (fasteners)

来源：评论

学校读者我要写书评

暂无评论

Combining N-gram retrieval with weights propagation on massive RDF graphs

Combining N-gram retrieval with weights propagation on massi...

引用

2012 9th International Conference on Fuzzy Systems and knowledge Discovery, FSKD 2012

作者： Hu, He Du, Xiaoyong Key Labs of Data Engineering and Knowledge Engineering Ministry of Education China School of Information Renmin University of China Beijing China

ISBN: (纸本)9781467300223

N-gram approach takes the position information into account additionally and thus can offer higher accuracy in query answering than keyword based approaches and is widely used in IR and NLP. However, in large-scale RDF graphs, URIs instead of documents are the ranking and querying units;URIs are usually much shorter than documents, and different URIs are interlinked into a massive network. One shot n-gram querying is usually not good for the RDF data in many cases. In this paper, we present a hybrid framework which combines the n-gram retrieval with link analysis based weight propagation. The idea is to exploit the link structures in the RDF data graphs and propagate the one shot n-gram score weights along with these links. Large scale experiments using MapReduce on Billion Triples Challenge dataset show the hybrid framework achieves an 80.3% improvement in relevance scores over mere n-gram retrieval. © 2012 IEEE.

关键词： MapReduce

来源：评论

学校读者我要写书评

暂无评论

Multi-Cluster Feature Selection Based on Isometric Mapping

引用

IEEE/CAA Journal of Automatica Sinica 2022年第3期9卷 570-572页

作者： Yadi Wang Zefeng Zhang Yinghao Lin Henan Key Laboratory of Big Data Analysis and Processing Henan UniversityKaifeng 475004 Institute of Data and Knowledge Engineering School of Computer and Information EngineeringHenan UniversityKaifeng 475004China

Dear editor,This letter presents an unsupervised feature selection method based on machine *** selection is an important component of artificial intelligence,machine learning,which can effectively solve the curse of d... 详细信息

关键词： problem letter dimensionality

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：