检索结果-内蒙古大学图书馆

A parallel scheduling algorithm for reinforcement learning in large state space

Frontiers of Computer Science 2012年第6期6卷 631-646页

作者： Quan LIU Xudong YANG Ling JING Jin LI Jiao LI Institute of Computer Science and Technology Soochow UniversitySuzhou 215006China Department of Computer Science and Technology Nanjing UniversityNanjing 210093China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Jilin UniversityChangchun 130012China

The main challenge in the area of reinforcement learning is scaling up to larger and more complex problems. Aiming at the scaling problem of reinforcement learning, a scalable reinforcement learning method, DCS-SRL, is proposed on the basis of divide-and-conquer strategy, and its convergence is proved. In this method, the learning problem in large state space or continuous state space is decomposed into multiple smaller subproblems. Given a specific learning algorithm, each subproblem can be solved independently with limited available resources. In the end, component solutions can be recombined to obtain the desired result. To ad- dress the question of prioritizing subproblems in the scheduler, a weighted priority scheduling algorithm is proposed. This scheduling algorithm ensures that computation is focused on regions of the problem space which are expected to be maximally productive. To expedite the learning process, a new parallel method, called DCS-SPRL, is derived from combining DCS-SRL with a parallel scheduling architecture. In the DCS-SPRL method, the subproblems will be distributed among processors that have the capacity to work in parallel. The experimental results show that learning based on DCS-SPRL has fast convergence speed and good scalability.

关键词： divide-and-conquer strategy parallel schedule scalability large state space continuous state space

来源：评论

学校读者我要写书评

暂无评论

Learning the structure of dynamic bayesian network with domain knowledge

Learning the structure of dynamic bayesian network with doma...

引用

2012 International Conference on Machine Learning and Cybernetics, ICMLC 2012

作者： Chen, Juan Jia, Haiyang Huang, Yuxiao Liu, Dayou College of Computer Science and Technology Key Laboratory for Symbolic Computation and Knowledge Engineering of Ministry of Education Jilin University Changchun 130012 China

ISBN: (纸本)9781467314855

Dynamic Bayesian Network (DBN) is a graphical model for representing temporal stochastic processes. Learning the structure of DBN is a fundamental step for parameter learning, inference and application. For large scale problem, the structure learning is intractable. In some domains the training data is very limited and noisy, so learning the DBN structure only with training data is impractical. Domain knowledge may improve both the efficiency and the accuracy of the learning algorithm. But usually, the domain knowledge is uncertainty, unclear and even with conflict. This paper presents a novel algorithm for learning the structure of DBN, which consider the data and domain knowledge simultaneously, empirical experiment shows that the proposed algorithm improved the efficiency and the accuracy of the DBN structure learning. © 2012 IEEE.

关键词： Bayesian networks

来源：评论

学校读者我要写书评

暂无评论

Quasi-periodic Solutions of the General Nonlinear Beam Equations

引用

Communications in Mathematical Research 2012年第1期28卷 51-64页

作者： GAO YI-XIAN College of Mathematics and Statistics Northeast Normal University Changchun 130024 Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Jilin University Changchun 130012

In this paper, one-dimensional （1D） nonlinear beam equations of the form utt - uxx ＋ uxxxx ＋ mu = f （u） with Dirichlet boundary conditions are considered, where the nonlinearity f is an analytic, odd function and f（u） = O（u3）. It is proved that for all m ∈ （0, M＊]  R （M＊ is a fixed large number）, but a set of small Lebesgue measure, the above equations admit small-amplitude quasi-periodic solutions corresponding to finite dimensional invariant tori for an associated infinite dimensional dynamical system. The proof is based on an infinite dimensional KAM theory and a partial Birkhoff normal form technique.

关键词： beam equation KAM theorem quasi-periodic solution partial Birkhoffnormal form

来源：评论

学校读者我要写书评

暂无评论

Research of the timer granularity based on linux

Research of the timer granularity based on linux

引用

International Conference on Computer, Informatics, Cybernetics and Applications 2011, CICA 2011

作者： Qingtian, Geng Nan, Jiang Hongwei, Zhao Junling, Liu Department of Computer Science and Technology Jilin University Changchun 130012 China Changchun Normal University Changchun 130032 China Key Laboratory of Symbolic Computation and Knowledge Engineering Ministry of Education Jilin University Changchun 130012 China

ISBN: (纸本)9789400718388

Time interrupts play an important role in the system. When time interrupts occur, system will inspect process running state, providing an opportunity to schedule, which is important for improving system real time performance. For the low system real-time ability of Linux, the paper adopt enhancing time granularity to improve system real time ability by means of one-shot triggering mode. The kernel of one-shot mode time interrupt is that: the system need not have to respond to any change every time, but only make response when having to respond. Experimental results show that the system real-time ability of Linux is improved through enhancing time granularity. © 2012 Springer Science+Business Media B.V.

关键词： Linux

来源：评论

学校读者我要写书评

暂无评论

A novel method for computing the inverse of cardinal direction relations

引用

Journal of Convergence Information Technology 2012年第1期7卷 130-138页

作者： Wang, Fang Liu, Dayou Dong, Yiqun College of Computer Science and Technology Jilin University Changchun China Key Laboratory of Symbolic Computation and Knowledge Engineering Ministry of Education

This paper focuses efforts on the problem of computing the inverse of cardinal direction relations (CRs), which is a fundamental problem in qualitative spatial reasoning. We study the quite expressive cardinal direction relation model for extended objects, known as cardinal direction calculus (CDC). We first concentrate on a set of a special type of CRs defined in CDC, named rectangle-CRs, and compute the inverse of rectangle-CRs by exploiting the evident connection between basic rectangle- CRs and interval relations. Then, we consider progressively the general cardinal direction relations in CDC, or called CDC relations for short, the inverse of which is computed by reducing to the computation of the inverse of rectangle-CRs. This simplifies the computations. Analyzing in theory and the final results both demonstrate that our algorithms are correct and complete and the time complexity is bounded by a constant number of operations.

关键词： Calculations

来源：评论

学校读者我要写书评

暂无评论

A collective method to predict missing links in networks

引用

Journal of computational Information Systems 2012年第13期8卷 5543-5552页

作者： Li, Lina Ouyang, Jihong Liu, Dayou Yang, Bo Chen, Huiling College of Computer Science and Technology Jilin University Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Changchun China

Link prediction in networks has attracted increasing attention by researchers coming from different branches of science, and has been applied in many important domains, such as protein-protein interaction networks in biology, friend-of-friend network in sociology. Many algorithms have been presented to solve this problem, but we find that these methods are attempting to predict link between pair of nodes separately, they assume that the existence of one link is independent with other ones. However, in most real world network, links may be interrelated. So in this paper, we introduce the collective idea, and present a collective inference framework that extends current prediction methods to estimate interrelated links together. Then two new algorithms are proposed based on this framework. As will be shown in the experimental section, we compare our approach with several existence methods on five networks, the results indicate that our method achieve better performance especially when few links are known. © 2011 by Binary Information Press.

关键词： Iterative methods

来源：评论

学校读者我要写书评

暂无评论

Conflict-based belief revision operators in possibilistic logic 26

Conflict-based belief revision operators in possibilistic lo...

引用

26th AAAI Conference on Artificial Intelligence and the 24th Innovative Applications of Artificial Intelligence Conference, AAAI-12 / IAAI-12

作者： Qi, Guilin Wang, Kewen Key Laboratory of Computer Network and Information Integration of State Education Ministry School of Computer Science and Engineering Southeast University China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Jilin University Changchun 130012 China School of Computing and Information Technology Griffith University Australia

ISBN: (纸本)9781577355687

In this paper, we investigate belief revision in possibilistic logic, which is a weighted logic proposed to deal with incomplete and uncertain information. Existing revision operators in possibilistic logic are restricted in the sense that the input information can only be a formula instead of a possibilistic knowledge base which is a set of weighted formulas. To break this restriction, we consider weighted prime implicants of a possibilistic knowledge base and use them to define novel revision operators in possibilistic logic. Intuitively, a weighted prime implicant of a possibilistic knowledge base is a logically weakest possibilistic term (i.e., a set of weighted literals) that can entail the knowledge base. We first show that the existing definition of a weighted prime implicant is problematic and need a modification. To define a revision operator using weighted prime implicants, we face two problems. The first problem is that we need to define the notion of a conflict set between two weighted prime implicants of two possibilistic knowledge bases to achieve minimal change. The second problem is that we need to define the disjunction of possibilistic terms. We solve these problems and define two conflict-based revision operators in possibilistic logic. We then adapt the well-known postulates for revision proposed by Katsuno and Mendelzon and show that our revision operators satisfy four of the basic adapted postulates and satisfy two others in some special cases. Copyright © 2012, Association for the Advancement of Artificial Intelligence. All rights reserved.

关键词： Problem solving

来源：评论

学校读者我要写书评

暂无评论

A random network ensemble model based generalized network community mining algorithm

引用

Zidonghua Xuebao/Acta Automatica Sinica 2012年第5期38卷 812-822页

作者： Yang, Bo Liu, Jie Liu, Da-You College of Computer Science and Technology Jilin University Changchun 130012 China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Jilin University Changchun 130012 China

According to the attributes of nodes and the linkages between them, most real-world complex networks could be assortative and disassortative. Community structures are ubiquitous in both types of networks. The ability to discovery meaningful community structures from both types of networks is fundamental for theoretical research and practical applications. Since the types of exploratory networks to be processed are usually unknown beforehand, it is difficult to determine what specific algorithms should be applied to them to obtain meaningful community structures. To address this issue, a novel concept of generalized network community is proposed in order to unify two concepts of assortative and disassortative communities. Based on a random network ensemble model, a generalized community mining algorithm, called G-NCMA, is proposed. Experimental results demonstrate that the G-NCMA algorithm is able to properly mine potential communities from explorative networks, as well as to determine their respective types. Copyright © 2012 Acta Automatica Sinica. All rights reserved.

关键词： Maximum likelihood estimation

来源：评论

学校读者我要写书评

暂无评论

SESSION SEGMENTATION METHOD BASED ON COBWEB

SESSION SEGMENTATION METHOD BASED ON COBWEB

引用

2012 IEEE 2nd International Conference on Cloud Computing and Intelligence Systems

作者： Zhenshan Hou Mingliang Cui Ping Li Liuliu Wei Wenhao Ying Wanli Zuo Key Laboratory of Symbolic Computation and Knowledge Engineering of the Ministry of Education College of Computer Science and TechnologyJilin University

Session segmentation can not only facilitate further study of users’ interest mining but also act as the foundation of other retrieval researches based on users’ complicated search behaviors. This paper proposes session boundary discrimination model （the binary classification tree） utilizing time interval and query likelihood on the basis of COBWEB. The model has prominently improved recall ratio, precision ratio and value F to more than 90 percent and particularly the value F for yes class rises compared with previous study. It is an incremental algorithm that can deal with large scale data, which will be perfectly applied into user interest mining. Owing to its good performance in session boundary discrimination, the application of the model can serve as a tool in fields like personalized information retrieval, query suggestion, search activity analysis and other fields which have connection with search results improvement.

关键词： Clustering COBWEB Query log session segmentation

来源：评论

学校读者我要写书评

暂无评论

A new regularization classification method based on extreme learning machine in network data

引用

Journal of Information and computational Science 2012年第12期9卷 3351-3363页

作者： Li, Lina Liu, Dayou Ouyang, Jihong Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education College of Computer Science and Technology Jilin University Changchun 130012 China

In this paper, we present a new regularization classification method based on extreme learning machine for within network node classification problem. In particular, we define a new objective function, which contains loss function, smoothness function, and intra-class regularization and inter-class regularization terms. In realization, we extend extreme learning machine to semi-supervised learning problem and deduce a new hidden layer weight, and then use the advanced extreme learning machine to optimize the objective function. Experiment results shows that our method can obtain higher accuracies than other methods when more than 30% nodes are labeled in the network. © 2012 Binary Information Press.

关键词： Supervised learning

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：