检索结果-内蒙古大学图书馆

International Conference on Audio, Language and Image processing, ICALIP

作者： Chi Zhang Xiaoqiang Li Wei Li Peizhong Lu Wenqiang Zhang School of Computer Engineering and Science Shanghai University Shanghai China Shanghai Key Laboratory of Intelligent Information Processing Fudan University Shanghai China School of Computer Science and Technology Fudan University Shanghai China

ISBN: (纸本)9781509006557

Speaker recognition in short speech condition is a difficult topic because the length of training and test speech is very short. One of the main disadvantage of the existing methods for speaker recognition is that they need very sufficient data and it's usually impossible in reality applications. In our experiments, the conventional methods with single feature don't make good performance in short speech. We propose a novel i-vector framework using multiple features and Principal Component Analysis (PCA) in short speech condition to overcome this difficulty, as multiple features combination can represent more aspects of a speaker. PCA is used to map the multiple features to an uncorrelated and orthogonal basis set to meet the requirements of Gaussian Mixture Model (GMM) with diagonal covariance matrices and i-vector. Improvement from the proposed approach compared to a state-of-the-art system are of roughly 50% relative at equal error rate when evaluated on the telephone conditions from the 2010 NIST speaker recognition evaluation (SRE).

关键词： Speech Feature extraction Principal component analysis Speaker recognition Covariance matrices Training Mel frequency cepstral coefficient

来源：评论

学校读者我要写书评

暂无评论

Neighborhood collaborative classifiers

Neighborhood collaborative classifiers

引用

2016 International Conference on Machine Learning and Cybernetics, ICMLC 2016

作者： Xu, Suping Yang, Xibei Tsang, Eric C. C. Mantey, Eric A. School of Computer Science and Engineering Jiangsu University of Science and Technology Zhenjiang212003 China School of Economics and Management Nanjing University of Science and Technology Nanjing210094 China Faculty of Information Technology Macau University of Science and Technology C-Macau519020 China Intelligent Information Processing Key Laboratory of Shanxi Province Shanxi University Taiyuan030006 China Key Laboratory of Oceanographic Big Data Mining and Application of Zhejiang Province Zhejiang Ocean University Zhoushan316022 China

ISBN: (纸本)9781509003891

In neighborhood rough set model, the majority rule based neighborhood classifier (NC) is easy to be misjudged with the increasing of the size of information granules. To remedy this deficiency, we propose a neighborhood collaborative classifier (NCC) based on the idea of collaborative representation based classification (CRC). NCC first performs feature selection with neighborhood rough set, and then instead of solving the classification problem by the majority rule, NCC solves a similar problem with collaborative representation among the neighbors of each unseen sample. Experiments on UCI data sets demonstrate that: 1) Our NCC achieves satisfactory performance in larger neighborhood information granules when compared with NC;2) NCC reduces the size of dictionary when compared with CRC. © 2016 IEEE.

关键词： Classification (of information)

来源：评论

学校读者我要写书评

暂无评论

Efficiency-Controllable Random Walks on a Class of Recursive Scale-Free Trees with a Deep Trap

引用

Chinese Physics Letters 2015年第3期32卷 13-16页

作者：李玲关佶红周水庚 Department of Computer Science and Technology Tongji University Shanghai 201804 Department of Basic Courses Zhejiang Shuren University Hangzhou 310015 School of Computer Science Fudan University Shanghai 200433 Shanghai Key Lab of Intelligent Information Processing Fudan University Shanghai 200433

Controls, especially effficiency controls on dynamical processes, have become major challenges in many complex systems. We study an important dynamical process, random walk, due to its wide range of applications for modeling the transporting or searching process. For lack of control methods for random walks in various structures, a control technique is presented for a class of weighted treelike scale-free networks with a deep trap at a hub node. The weighted networks are obtained from original models by introducing a weight parameter. We compute analytically the mean first passage time （MFPT） as an indicator for quantitatively measurinM the et^ciency of the random walk process. The results show that the MFPT increases exponentially with the network size, and the exponent varies with the weight parameter. The MFPT, therefore, can be controlled by the weight parameter to behave superlinearly, linearly, or sublinearly with the system size. This work provides further useful insights into controllinM eftlciency in scale-free complex networks.

关键词： Efficiency-Controllable Random Walks on a Class of Recursive Scale-Free Trees with a Deep Trap

来源：评论

学校读者我要写书评

暂无评论

Exploiting Local and Global Characteristics for Contrast Based Visual Saliency Detection

引用

Journal of shanghai Jiaotong university(science) 2015年第1期20卷 14-20页

作者：徐新王英林张晓龙 School of Computer Science and Technology Wuhan University of Science and Technology Hubei Province Key Laboratory of Intelligent Information Processing and Real-time Industrial System Wuhan University of Science and Technology School of Information Management and Engineering Shanghai University of Finance and Economics

Visual saliency is an important cue in human visual system to identify salient region in the image;it can be useful in many applications including image retrieval,object recognition,image segmentation,*** contrast has been used as an effective feature to detect visual salient ***,the conventional contrast measures either in spectral domain or in spatial domain fail to give sufficient consideration towards the local and global characteristics of the *** paper presents a visual saliency detection algorithm based on a novel contrast *** measurement extracts the spectral information of image block using the 2D discrete Fourier transform(DFT),and combines with the total variation(TV)of image block in spatial *** proposed algorithm is used to perform salient region detection in the image,and compared with state-of-the-art *** experimental results from the MSRA dataset validate the effectiveness of the proposed algorithm.

关键词： visual saliency contrast measure multi-scale local contrast global contrast

来源：评论

学校读者我要写书评

暂无评论

EXPRESS WAYBILL EXTRACTION ON PARCEL IMAGES BY LOGO MATCHING

EXPRESS WAYBILL EXTRACTION ON PARCEL IMAGES BY LOGO MATCHING

引用

IEEE International Conference on Image processing

作者： Shujing Lu Jie Zhao Yue Lu ECNU-SRI Joint Lab for Pattern Analysis and Intelligent System Shanghai Research Institute of China Post Group Shanghai Key Laboratory of Multidimensional Information Processing Department of Computer Science and Technology East China Normal University

ISBN: (纸本)9781467399623

We propose an efficient method for automatically extracting express waybills from parcel images, which is challenging due to varied resolution of parcel images, arbitrary direction of waybills and different information filled by senders. To address these challenges, logo matching is employed to extract the waybills. We begin by extracting scale-invariant feature-transformation (SIFT) keypoints from both the reference logo image and a parcel image, and matching them subject to a consistent projective transformation (homography) by using random sample consensus (RANSAC). Once the homography matrix is computed, we extract the waybill of parcel image by mapping all pixels from a standard waybill image to the parcel image. Experimental results on test datasets demonstrate the effectiveness of the proposed method.

关键词： Logo matching Express waybill extraction SIFT images Sifting Fields LOGO Imagery (Psychotherapy) Image homography

来源：评论

学校读者我要写书评

暂无评论

Enhancing Time Series Clustering by Incorporating Multiple Distance Measures with Semi-Supervised Learning

引用

Journal of computer science & technology 2015年第4期30卷 859-873页

作者：周竞朱山风黄晓地张彦春 School of Computer Science Fudan University Shanghai 200433 China Shanghai Key Laboratory of Intelligent Information Processing Fudan University Shanghai 200433 China School of Computing and Mathematics Charles Sturt University Albury NSW 2640 Australia School of Engineering and Science Victoria University Melbourne Victoria 8001 Australia Shanghai Key Laboratory of Data Science Fudan University Shanghai 201203 China

Time series clustering is widely applied in various areas. Existing researches focus mainly on distance measures between two time series, such as dynamic time warping （DTW） based methods, edit-distance based methods, and shapelets-based methods. In this work, we experimentally demonstrate, for the first time, that no single distance measure performs significantly better than others on clustering datasets of time series where spectral clustering is used. As such, a question arises as to how to choose an appropriate measure for a given dataset of time series. To answer this question, we propose an integration scheme that incorporates multiple distance measures using semi-supervised clustering. Our approach is able to integrate all the measures by extracting valuable underlying information for the clustering. To the best of our knowledge, this work demonstrates for the first time that the semi-supervised clustering method based on constraints is able to enhance time series clustering by combining multiple distance measures. Having tested on clustering various time series datasets, we show that our method outperforms individual measures, as well as typical integration approaches.

关键词： time series analysis clustering dynamic programming information search and retrieval

来源：评论

学校读者我要写书评

暂无评论

Hybrid discrimination method for samples classification in medicine

引用

International Journal of Simulation: Systems, science and technology 2016年第27期17卷 17.1-17.9页

作者： Chen, Xiaoyu Liu, Bo Xia, Xin Yan, Dandan Yan, Wang Ma, Lizhuang Department of Information Center East Hospital Tongji University School of Medicine Jimo Road 150 Shanghai China School of Continuing Education Shanghai Jiao Tong University Shanghai200240 China Provincial Key Laboratory for Computer Information Processing Technology Soochow University Suzhou215006 China Department of Computer Science Shanghai Jiao Tong University Shanghai200240 China

As indispensable solutions in classification problems, discrimination for samples has been employed in medicine, and it is performed subjectively by physicians at present, which hinders the diagnosis and treatment in medicine. In this paper, a hybrid discrimination method (HDM) in medicine is proposed, which consists of two phases, including attribute selection phase and discriminant phase. In attribute selection phase, critical attributes are selected from the original features by linear correlation and C5.0 decision tree. In discriminant phase, samples are discriminated by discriminant analysis. This discrimination method is evaluated through five datasets of chronic hepatitis B, cardiac Single Proton Emission Computed Tomography (SPECT) images, Lung Cancer, Hepatitis survival and Iris plant for demonstrating its viability and applications. Finally, this proposed method has obtained the critical clinical lab indicators and discriminants related to three syndromes in CHB dataset, and it also performs well than some typical classification methods in the other four datasets for its broader applications. © 2016, UK Simulation Society. All rights reserved.

关键词： Discriminant analysis

来源：评论

学校读者我要写书评

暂无评论

Improved protein residue-residue contacts prediction using learning-to-rank

Improved protein residue-residue contacts prediction using l...

引用

IEEE International Conference on Bioinformatics and Biomedicine (BIBM)

作者： Xiaoyang Jing Qiwen Dong School of Computer Science Fudan University Shanghai People's Republic of China Shanghai Key Laboratory of Intelligent Information Processing Shanghai People's Republic of China Key Laboratory of Network Oriented Intelligent Computation Harbin Institute of Technology Shenzhen Graduate School Shenzhen Guangdong People's Republic of China Institute for Data Science and Engineering East China Normal University Shanghai People's Republic of China

ISBN: (纸本)9781509016129

Protein residue-residue contacts dictate the topology of protein structure and play an important role in structural biology, especially in de novo protein structure prediction. Accurate prediction of residue contacts could improve the performance of de novo protein structure prediction methods. In this study, a novel method based on learning-to-rank (RRCRank) has been presented to predict protein residue-residue contacts. The proposed method formulates the contacts prediction problem as a ranking problem. Firstly, the contact probabilities of residue pairs are predicted by ensemble machine-learning classifiers and correlated mutations approaches. And then, the proposed method integrates the complementary outputs of machine-learning and correlated mutations approaches and uses the learning-to-rank algorithm to rank residue pairs based on their probabilities to be contacts. Benchmarked on the CASP11 dataset, the proposed method achieves an improved performance for all three categories of contacts (short-range, medium-range and long-range contacts), which shows the proposed method based on learning-to-rank could take advantage of machine-learning and correlated mutations approaches and could provide the state-of-the-art performance.

关键词： Proteins Protein engineering Training Predictive models Feature extraction Classification algorithms Prediction algorithms

来源：评论

学校读者我要写书评

暂无评论

Online hot event discovery based on Association Link Network

Online hot event discovery based on Association Link Network

引用

作者： Liu, Yang Luo, Xiangfeng Xuan, Junyu School of Computer Engineering and Science Shanghai University China Key Laboratory of Intelligent Information Processing Institute of Computing Technology Chinese Academy of Science Beijing China

Online hot event discovery has become a flourishing frontier where online document streams are monitored to discover newly occurring events or assigned to previously detected events. However, hot events have the nature to evolve, and their inherent topically related words are also likely to evolve. It makes event discovery a challenging task for traditional-mining approaches. Combining word association and semantic community, Association Link Network (ALN) organizes the loosely distributed associated resources. This paper presents an ALN-based novel online hot event discovery approach. Technically, this approach is enacted around three stages. In the first stage, we extract significant features to represent the content of each document from the online document stream. During the second stage, we classify the online document stream into topically related detected events considering event evolution in the form of ALN. At the third stage, we create an ALN-based event detection algorithm, which is used to timely discover newly occurring hot events. The online datasets used in our empirical studies are acquired from Baidu News, which spans a range of 1315 hot events and 236,300 documents. Experimental results demonstrate the hot events discovery ability with respect to high accuracy, good scalability, and short runtime. © 2015 John Wiley & Sons, Ltd.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

Efficient Variational Light Field View Synthesis for Making Stereoscopic 3D Images

Efficient Variational Light Field View Synthesis for Making ...

引用

作者： Zhang, Lei Zhang, Yu-Hang Huang, Hua Beijing Key Lab of Intelligent Information Technology School of Computer Science Beijing Institute of Technology China

We present a novel approach for making stereoscopic images by variational view synthesis on the multi-perspective light field. With the intended disparities as constraints, we specialize the generative variational model by incorporating per-pixel viewpoint assignment to synthesize the stereo pair. Also, we improve the variational solution by use of explicit weighted average on the light field. Our algorithm is able to handle arbitrary disparity remapping, thus enabling more flexible disparity control for the desired stereoscopic effect. The experiments demonstrate the effectiveness and efficiency for making the stereoscopic 3D images based on the light field. © 2015 The Author(s) computer Graphics Forum.

关键词： Stereo image processing

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：