检索结果-内蒙古大学图书馆

International Conference on Transportation, Mechanical, and Electrical Engineering (TMEE)

作者： Xinfu Li Wenxue Cao Key Laboratory of Machine Learning and Computational Intelligence College of Mathematics and Computer Science Hebei University Baoding China

The phenomenon of person name ambiguity is widespread on web pages in that one name may be used by different people. It is important to uniquely identify the given person on the web. In this paper, the method Baidu-PND is proposed by the authors. It is an unsupervised name disambiguation method based on Baidu Encyclopedia. We extract three features including background knowledge, contextual feature and Related-Set of the characters from the online Baidu Encyclopedia. The weights of the features are studied by logistic regression algorithm. Then we make a linear fusion of the features. The maximum combined value is selected as the correct person on web pages. Experiments are conducted to measure the performance of Baidu-PND, which show that the performance is higher than we expected, validating its feasibility and effectiveness for person name disambiguation on web pages. And, Baidu-PND is a new method for knowledge mining based on Baidu Encyclopedia.

关键词： Encyclopedias Web pages Feature extraction Context Physics Accuracy Educational institutions

来源：评论

学校读者我要写书评

暂无评论

L1-Norm-Based 2DLPP

L1-Norm-Based 2DLPP

引用

2011 China Control and Decision Conference(2011中国控制与决策会议 CCDC)

作者： Hao-Xin Zhao Hong-Jie Xing Xi-Zhao Wang Jun-Fen Chen Key Laboratory of Machine Learning and Computational Intelligence College of Mathematics and Computer Science Hebei University Baoding 071002 China

In this paper, we propose a new L1-Norm-Based two-dimensional locality preserving projections (2DLPP-L1). Traditional 2D-LPP can preserve local structure and extract feature directly form matrices, which shows great advantages. However, it is based on L2 norm. It is well known that L2-norm-based criterion is sensitive to outliers. We generalize 2D-LPP to its corresponding L1-norm-based version, i.e. 2DLPP-L1, which is more robust against outliers. To evaluate the performance of 2DLPP-L1, several experiments are performed on the ORL face databases. Experimental results demonstrate that 2DLPP-L1 has better performance than its related methods.

关键词： L1 norm 2DLPP outliers two dimensional projections

来源：评论

学校读者我要写书评

暂无评论

Regional objects based image retrieval

Regional objects based image retrieval

引用

2011 China Control and Decision Conference(2011中国控制与决策会议 CCDC)

作者： Jian-Guo Wu Xi-Zhao Wang Hong-Jie Xing Key Laboratory of Machine Learning and Computational Intelligence College of Mathematics and Computer Science Hebei University Baoding 071002 China

Content-based image retrieval has become an important research area. In order to extract the semantic information within the user’s query concept, we propose an image retrieval method based on regional objects. It is regarded as the pre-processing of a given query image, that is to say, when we get a query image, it needs us to segment the regional object which is useful or interesting, and retrieve according to the segmented fragment. Moreover, we propose a correlation coefficient based color representation. Experimental results demonstrate that our proposed approach performs much better than its related methods. Furthermore, the presented system has a high retrieval precision and keeps color consistency between the similarity images.

关键词： Content-based image retrieval Regional objects Pre-processing Correlation coefficient

来源：评论

学校读者我要写书评

暂无评论

A kernel two-sample test

The Journal of Machine Learning Research

引用

The Journal of machine learning Research 2012年第1期13卷

作者： Arthur Gretton Karsten M. Borgwardt Malte J. Rasch Bernhard Schölkopf Alexander Smola MPI for Intelligent Systems Tübingen Germany Machine Learning and Computational Biology Research Group Max Planck Institutes Tübingen Tübingen Germany State Key Laboratory of Cognitive Neuroscience and Learning Beijing Normal University Beijing P.R. China Yahoo! Research Santa Clara CA and The Australian National University Canberra ACT Australia

We propose a framework for analyzing and comparing distributions, which we use to construct statistical tests to determine if two samples are drawn from different distributions. Our test statistic is the largest difference in expectations over functions in the unit ball of a reproducing kernel Hilbert space (RKHS), and is called the maximum mean discrepancy (MMD).We present two distribution free tests based on large deviation bounds for the MMD, and a third test based on the asymptotic distribution of this statistic. The MMD can be computed in quadratic time, although efficient linear time approximations are available. Our statistic is an instance of an integral probability metric, and various classical metrics on distributions are obtained when alternative function classes are used in place of an RKHS. We apply our two-sample tests to a variety of problems, including attribute matching for databases using the Hungarian marriage method, where they perform strongly. Excellent performance is also obtained when comparing distributions over graphs, for which these are the first such tests.

关键词： hypothesis testing integral probability metric kernel methods schema matching two-sample test uniform convergence bounds

来源：评论

学校读者我要写书评

暂无评论

A survey of the initialization of centers and widths in radial basis function network for classification

A survey of the initialization of centers and widths in radi...

引用

International Conference on machine learning and Cybernetics (ICMLC)

作者： Chun-Ru Dong Patrick P. K. Chan Wing W. Y. Ng Daniel S. Yeung Key Laboratory of Machine Learning and Computational Intelligence College of Mathematics and Computer Science Hebei University Baoding China Machine Learning and Cybernetics Research Center School of Computer Science and Engineering South China University of Technology Guangzhou China

The radial basis function network (RBFN) has been widely used in various fields such as function regression, pattern recognition, and error detection, etc. However, the structural parameters of RBFN including the number of hidden units, centers vectors, and widths (variances) are one of the most important issues when training a RBFN, which greatly affect the performance of RBFN. So, the objective of this paper is to construct an elementary survey about this problem. Firstly, the fundamental knowledge and notations of RBFN is introduced. Secondly, we summarize most existing network structure initialization methods for RBFN and categorize them into four goups. Then some typical appraoches for each category are introduced and discussed. The disadvantages and virtues for parts of methods are also introduced. Finally, the paper is concluded with a discussion of current difficulties and possible future directions about RBFN architecture selection.

关键词： Training Neurons Artificial neural networks Clustering algorithms Optimization machine learning

来源：评论

学校读者我要写书评

暂无评论

Chinese keyword search over relational databases

Chinese keyword search over relational databases

引用

World Congress on Software Engineering

作者： Zhu, Liang Zhu, Yong Ma, Qin Key Laboratory of Machine Learning and Computational Intelligence School of Mathematics and Computer Science Hebei University Baoding Hebei 071002 China Department of Foreign Language Teaching and Research Hebei University Baoding Hebei 071002 China

ISBN: (纸本)9780769543031

Based on a knowledge base, we propose a new method to realize free-style Chinese keyword search over relational databases. Firstly, an index (also called knowledge base) is built by extracting related information of Chinese tuple words in a database, then query words and tuple words are matched quickly each other by with the index to obtain some of candidate answers for a given query. Secondly, we present a new ranking strategy to compute similarities between the query and the candidate answers, and refine the candidate answers by a matching algorithm proposed in this paper. Finally, we rank refined candidate answers to obtain top-N results. The experiments are conducted using a real dataset, and the experimental results show that our method is efficient and effective. © 2010 IEEE.

关键词： Search engines

来源：评论

学校读者我要写书评

暂无评论

Layout identification of printed mathematical formula for recognition

Layout identification of printed mathematical formula for re...

引用

International Conference on Information Engineering and Computer Science

作者： Tian, Xue-Dong Zhao, Yan Wang, Hui Wang, Qiang-Jun College of Mathematics and Computer Science Hebei University Baoding China Hebei Key Laboratory of Machine Learning and Computational Intelligence Baoding China Baoding Senior Technical College Baoding China College of Literature Hebei University Baoding China

ISBN: (纸本)9781424479412

Printed mathematical formulas edited by different soft wares have some obvious differences. To distinguish it before recognition is beneficial to the formula recognition. Based on the statistical analysis to the characteristics of Latex and Word typesetting, an algorithm is designed to identify Latex and Word layout of printed formulas. The experiment indicates that this algorithm can achieve a relative high accuracy to standard printed documents. ©2010 IEEE.

关键词： Latexes

来源：评论

学校读者我要写书评

暂无评论

The application of decision tree in Chinese email classification

The application of decision tree in Chinese email classifica...

引用

International Conference on machine learning and Cybernetics (ICMLC)

作者： Hao Chen Yan Zhan Yan Li Key Laboratory of Machine Learning and Computational Intelligence College of Mathematics and Computer Science Hebei University Baoding China

Email is a kind of semi-structured document, some important attributes are contained in its structure, and especially using spam-specific features could improve the email classification results. In this paper, we apply decision tree data mining technique to dig out the potential association rules among these attributes of email, and then to identify unknown email's category based on these rules. According to the experiment of applying numerous Chinese emails to our email classifier, the efficiency of our method is not lower than that of other existing methods of checking whole email content text. Meanwhile our method can reduce the cost of computation and consumption of system resources.

关键词： Electronic mail Postal services Classification algorithms machine learning Classification tree analysis Association rules

来源：评论

学校读者我要写书评

暂无评论

A survey on active learning strategy

A survey on active learning strategy

引用

International Conference on machine learning and Cybernetics (ICMLC)

作者： Li-Li Sun Xi-Zhao Wang Key Laboratory of Machine Learning and Computational Intelligence College of Mathematics and Computer Science Hebei University Baoding China

Active learning is a hot topic in machine learning field. The main task of active learning is to automatically select the representative instances for efficiently reducing the sample complexity. This paper presents a brief survey of active learning regarding selection methods, query strategies, applications and other related works.

关键词： machine learning Training Classification algorithms Uncertainty learning systems Complexity theory Support vector machines

来源：评论

学校读者我要写书评

暂无评论

Support vector machine based on a new reduced samples method

Support vector machine based on a new reduced samples method

引用

International Conference on machine learning and Cybernetics (ICMLC)

作者： Shu-Xia Lu Jie Meng Gui-En Cao Key Laboratory of Machine Learning and Computational Intelligence College of Mathematics and Computer Science Hebei University Baoding China

The support vectors play an important role in the training to find the optimal hyper-plane. For the problem of many non-support vectors and a few support vectors in the classification of SVM, a method to reduce the samples that may be not support vectors is proposed in this paper. First, adopt the Support Vector Domain Description to find the smallest sphere containing the most data points, and then remove the objects outside the sphere. Second, remove the edge points based on the distance of each pattern to the centers of other classes. In comparison with the standard SVM, the experimental results show that the new algorithm in the paper is capable of reducing the number of samples as well as the training time while maintaining high accuracy.

关键词： Training Accuracy Classification algorithms Kernel Support vector machine classification machine 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：