检索结果-内蒙古大学图书馆

2008 International Conference on Machine Learning and Cybernetics(2008机器学习与控制论国际会议)

作者： NENG-HAI YU LIAN-SHENG ZHUANG WEI ZHOU MOE-Microsoft Key Laboratory of Multimedia Computing and Communication University of Science and Te School of Information Science and Technology University of Science and Technology of China Hefei

Most methods for multiclass objects learning have large computational complexity and samples scale complexity. In this paper, within the framework of boosting, we propose a novel method called JointBoosting-GA. It is suitable to all datasets from small to very large, and results in a much faster classifier at run time. To achieve it, we combine two ideas: 1) Firstly, we introduce a novel technique, which is based on genetic algorithm, to generate new samples. At each boosting round, it generates new samples and expands the training set. By this way, our method can avoid overfitting, and produce classifiers with high predictive accuracy. 2) Secondly, by sharing features across classes, we reduce the computational cost of the learned classifiers at run time, when detecting multiclass objects in cluttered scenes. Experiments on Caltech 101 dataset showed that, our method outperformed SVM and JointBoosting when only small samples were available for multiclass objects learning.

关键词： Multiclass learning Boosting Genetic algorithm Shared feature

来源：评论

学校读者我要写书评

暂无评论

Doctored JPEG image detection

Doctored JPEG image detection

引用

IEEE International Conference on multimedia and Expo (ICME)

作者： Weihai Li Nenghai Yu Yuan Yuan MOE-Microsoft Key Laboratory of Multimedia Computing and Communication University of Science and Technology Hefei China School of Engineering and Applied Science Aston University Birmingham UK

Nowadays, digital images can be easily modified by using software. In this paper, a new blind approach is proposed to detect copy-paste trail in a doctored JPEG image, i.e., to check whether a copied area came from the same image or not. When a copy-paste procedure is done on an image, especially adding or hiding an object, the block artifact grid contained in the copy-pasted slice is moved together. Since a slice must be placed properly in the target image to avoid obvious vision flaw, the grid in the slice mismatches to the original grid in the target image normally. Our approach utilizes the mismatch information of block artifact grid as a clue of copy-paste forgery. Experiment results demonstrate the efficiency of the proposed approach.

关键词： Transform coding Medical services Forgery Discrete cosine transforms Image coding Algorithm design and analysis Barium

来源：评论

学校读者我要写书评

暂无评论

Peak detection using peak tree approach for mass spectrometry data

引用

International Journal of Hybrid Intelligent Systems 2008年第4期5卷 197-208页

作者： Zhang, Peng Li, Houqiang Zhou, Xiaobo Wong, Stephen MOE-Microsoft Key Laboratory of Multimedia Computing and Communication University of Science and Technology of China Department of Radiology & Bioinformatics Core The Methodist Hospital Research Institute Weill Cornell Medical College Houston TX USA

In mass spectrometry (ms) analysis, false peak detection results are unavoidable due to severe spectrum variations. However, most current peak detection methods are neither robust enough to resist spectrum variations nor flexible enough to revise false detection results. To solve these two problems, we first propose peak tree to reveal the hierarchical relation among peak judgments made on different scales. Different peak tree decomposition will lead to different peak detection result, which make it very convenient to revise false result. Then, we propose a closed-loop scheme to iteratively refine peak tree decomposition. Experiment results show that, compared with conventional peak detection methods, our method can better resist spectrum variations and provide a more consistent result among different spectra.

关键词： Mass spectrometry peak detection wavelet scale space theory peak tree

来源：评论

学校读者我要写书评

暂无评论

A Joint Appearance-Spatial Distance for Kernel-Based Image Categorization

A Joint Appearance-Spatial Distance for Kernel-Based Image C...

引用

26th IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2008), vol.1

作者： Guo-Jun Qi Xian-Sheng Hua Yong Rui Jinhui Tang Zheng-Jun Zha Hong-Jiang Zhang MOE-Microsoft Key Laboratory of Multimedia Computing and Communication & Department of Automation University of Science and Technology Hefei Anhui China Microsoft Research Asia Beijing China

The goal of image categorization is to classify a collection of unlabeled images into a set of predefined classes to support semantic-level image retrieval. The distance measures used in most existing approaches either ignored the spatial structures or used them in a separate step. As a result, these distance measures achieved only limited success. To address these difficulties, in this paper, we propose a new distance measure that integrates joint appearance-spatial image features. Such a distance measure is computed as an upper bound of an information-theoretic discrimination, and can be computed efficiently in a recursive formulation that scales well to image size. In addition, the upper bound approximation can be further tightened via adaption learning from a universal reference model. Extensive experiments on two widely-used data sets show that the proposed approach significantly outperforms the state-of-the-art approaches.

关键词： Asia Image retrieval Upper bound Spatial coherence multimedia computing Automation Size measurement Kernel Design methodology Embedded computing

来源：评论

学校读者我要写书评

暂无评论

Learning object from small and imbalanced dataset with Boost-BFKO

Learning object from small and imbalanced dataset with Boost...

引用

IEEE International Conference on multimedia and Expo (ICME)

作者： Liansheng Zhuang Wei Zhou Qi Tian Nenghai Yu MOE-Microsoft Key Laboratory of Multimedia Computing and Communication University of Science and Technology Hefei China Department of Computer Science University of Texas San Antonio TX USA

One of the main drawbacks of boosting is its overfitting and poor predictive accuracy when the training dataset is small and imbalanced. In this paper, we introduce a novel learning algorithm Boost-BFKO, which combines boosting and data generation. It is suitable for small and imbalanced training datasets. To enlarge training sets, Boost-BFKO uses the adaptive Balanced Feature Knockout procedure (BFKO) to generate new synthetic samples. To enrich the training sets, Boost-BFKO selects seed samples from the minority class, and rebalances the total weights of the different classes in the updated training dataset. Experiments on Caltech 101 database showed that our method achieves a desirable performance when only a few training samples are available for binary classification and multiple object classification.

关键词： Classification algorithms Training Boosting Support vector machines Prediction algorithms Training data Accuracy

来源：评论

学校读者我要写书评

暂无评论

Boosting boostrap FLD subspaces for multiclass problem

Boosting boostrap FLD subspaces for multiclass problem

引用

MIPPR 2007: Pattern Recognition and Computer Vision

作者： Tuo, Wang Daoyi, Shen Lei, Wang Nenghai, Yu Nation's MOE-MS Key Laboratory of Multimedia Computing and Communication University of Science and Technology of China Hefei 230027 China

ISBN: (纸本)9780819469526

In this paper an ensemble feature extraction algorithm is proposed based on Adaboost.M2 for multiclass classification problem. The proposed algorithm makes no assumption about the distribution of the data and primarily performs by directly selecting the discriminant features with the minimum training error, which can overcome the main drawbacks of the traditional methods, such as Principle Component Analysis (PCA), Fisher Linear Discriminant Analysis (FLD) and Nonparametric Discriminant Analysis (NDA). The proposed method first samples a large number of bootstrap training subsets from the original training set and implements FLD in each subset to get a large number of bootstrap FLD projections. Then at each step of Adaboost.M2, the projection with the minimum weighted K Nearest Neighbor (KNN) classification error is selected from a pool of linear projections to combine the final strong classifier. Experimental results on three real-world data sets demonstrate that the proposed algorithm is superior to other feature extraction techniques.

关键词： Feature extraction

来源：评论

学校读者我要写书评

暂无评论

Parzen windows estimation using laplace kernel: A novel parametric analysis with information content

Parzen windows estimation using laplace kernel: A novel para...

引用

8th ACIS International Conference on Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed computing/3rd ACIS International Workshop on Self-Assembling Wireless Networks

作者： He, Jingsong Tang, Jian Fang, QianSheng MOE-Micro Soft Key Laboratory of Multimedia Computing and Communication Department of Electronic Science and Technology University of Science and Technology of China Anhui Institute of Architecture and Industry China

ISBN: (纸本)0769529097

Parzen windows estimation is one of the classical non-parametric methods in the field of machine learning and pattern classification, and usually uses Gaussian density function as the kernel. Although the relation between the kernel density estimation (KDE) and low-pass filtering is well known, it is vary difficult to setting the parameters of the other kinds of density, functions. This paper proposes a novel method to deal with the parameters of Laplace kernel through measuring the degree of exchanged information among interpolating points. Experimental results showed that the proposed method can improve the performance of Parzen windows significantly.

关键词： Learning systems

来源：评论

学校读者我要写书评

暂无评论

Anisotropic manifold ranking for video annotation

Anisotropic manifold ranking for video annotation

引用

IEEE International Conference onmultimedia and Expo, ICME 2007

作者： Tang, Jinhui Hua, Xian-Sheng Qi, Guo-Jun Mei, Tao Wu, Xiuqing MOE-Microsoft Key Laboratory of Multimedia Computing and Communication University of Science and Technology of China Hefei 230027 China Microsoft Research Asia Beijing 100080 China

ISBN: (纸本)1424410177

Graph-based semi-supervised learning (SSL) has attracted lots of interests in machine learning community as well as many application areas including video annotation recently. However, one of the two basic assumptions, structure assumption, which is an essential point of graph-based SSL, is not embedded into the pairwise similarity measure. Accordingly, we propose a novel graph-based SSL method for video annotation, named Anisotropic Manifold Ranking (AniMR), based on a structure-related similarity measure. This method takes the influence of the density difference between samples into account to improve the pairwise similarity. Furthermore, we will show that AniMR can also be deduced from partial differential equation (PDE) based anisotropic diffusion. It demonstrates that the label propagation in AniMR is anisotropic, which is intrinsically different from the isotropic label propagation process in general graph-based SSL methods. Experiments conducted on the TRECVID data set show this approach outperforms ordinary graph-based SSL methods and is effective for video semantic annotation. ©2007 IEEE.

关键词： Anisotropy

来源：评论

学校读者我要写书评

暂无评论

Study on the property of training samples and learning space with genetic algorithms

Study on the property of training samples and learning space...

引用

SNPD 2007: 8th ACIS International Conference on Software Engineering, Artificial Intelligence, Networking, and Parallel/Distributed computing

作者： He, Jingsong Tang, Jian Fang, QianSheng MOE-Microsoft Key Laboratory of Multimedia Computing and Communication Department of Electronic Science and Technology University of Science and Technology of China Anhui Institute of Architecture and Industry China

ISBN: (纸本)0769529097

Historically, the empirical risk of a pattern classifier was asked to be made zero, therefor the default property of training samples were limited to a separable ones. Nowadays on the contrary, the major idea of learning classification no longer ask the empirical risk of classifier must be made zero. In this situation, inseparable feature sets may not be detrimental to the performance of classifier. However, so far no experimental studies and analytical results show whether an inseparable feature set is available or not. This paper firstly analyzes the interaction between learning algorithms and feature selection, and gives a proof by both the analytical analysis and experimental studies. © 2007 IEEE.

关键词： Learning systems

来源：评论

学校读者我要写书评

暂无评论

A novel approach for signal transduction networks simulation at a mesoscopic level

A novel approach for signal transduction networks simulation...

引用

7th IEEE International Conference on Bioinformatics and Bioengineering, BIBE

作者： Shao, Chenxi Deng, Hongli MOE-Microsoft Key Laboratory of Multimedia Computing and Communication University of Science and Technology of China Hefei Anhui 230027 China Department of Computer Science and Technology University of Science and Technology of China Hefei Anhui 230027 China Anhui Province Key Laboratory of Software in Computing and Communication Hefei Anhui 230027 China

ISBN: (纸本)1424415098

Signal transduction (ST) networks simulation is important to medical research. However, owing to the complexity of the networks, most methods presented for the simulation are not desirable. Here, based on multi-agent system (MAS), we proposed a novel approach characterized by a multi-level discrete space and mesoscopic molecule-agents for describing ST networks at a mesoscopic level. The space is developed for the representation of the cellular space;it provides a relatively sufficient spatial representation with less computational cost. The mesoscopic molecule-agents model the active components within the networks, which will offer new potentials to overcome the combinatorial complexity as well as make computation feasible. To validate this method, the simulations of a composition/decomposition system and the Mitogen-Activated Protein Kinase (MAPK) signal transduction pathway are implemented. Finally, an in-silico experiment with perturbation in the MAPK pathway is made. The results obtained from the simulations demonstrate the validity of the approach. ©2007 IEEE.

关键词： Molecules

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：