检索结果-内蒙古大学图书馆

A new statistics-based evaluation measure for classification models

Journal of computational Information Systems 2014年第20期10卷 8775-8782页

作者： Ding, Tienan Li, Xiongfei Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Jilin University Changchun China College of Computer Science and Technology Jilin University Changchun China Pinggang Middle School Shenzhen China

In this paper, we investigate the problem of evaluating the performance of classification models. First of all we propose the concept of weighted correct pair map. Then based on the weighted correct pair map, we proposed a new evaluation measure. The attractive features of the measure are that it is insensitive to imbalanced class distributions and discriminating enough. Experimental results demonstrate that the proposed measure is reliable. The work presented in this paper may stimulate new research in classification model designing, such as designing new optimization-based classification or ranking models.

关键词： Information systems

来源：评论

学校读者我要写书评

暂无评论

Iris recognition based on probabilistic features

引用

Journal of computational Information Systems 2014年第20期10卷 8783-8792页

作者： Liu, Zhaojun Li, Xiongfei Zhao, Zhenting Key Laboratory of Symbolic Computation and Knowledge Engineering for Ministry of Education Jilin University Changchun China College of Computer Science and Technology Jilin University Changchun China College of Software Jilin University Changchun China

In this paper, we propose an iris recognition algorithm based on probabilistic features. At the image preprocessing step, we do many experiments on iris segmentation in order to minimize the negative influence on recognition caused by eyelash and eyelid. As different segmentation angle has different results, we have determined an ideal angle for further recognition. At the recognition step, iris is transformed into frequency domain by using Log Gabor filters. Then raw features are extracted. Probabilistic features, with two forms of distribution, random distribution and normal distribution, are set of vectors generated from these raw features. At the matching step, the raw features are replaced by the property features. KNN algorithm is used to decide the final result. Experiment results showed that the proposed method can make further improvement of recognition rates and reduce the time consumption significantly.

关键词： Normal distribution

来源：评论

学校读者我要写书评

暂无评论

A new epistasis detecting algorithm based on ant colony optimization 14

A new epistasis detecting algorithm based on ant colony opti...

引用

6th International Conference on Internet Multimedia Computing and Service, ICIMCS 2014

作者： Zhou, Zhihui Liu, Guixia Su, Lingtao Han, Liang Yan, Lun College of Computer Science and Technology Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Jilin University Changchun 130012 China

ISBN: (纸本)9781450328104

The rapid developments of chip-based technology have greatly improved human genetics and made routine the access of thousands of single nucleotide polymorphisms (SNPs) contributing to an informatics challenge. The characterization and interpretation of genes and gene-gene interactions that affect the susceptibility of common, complex multifactorial diseases is a computational and statistical challenge in genome-wide association studies (GWAS). Various methods have been proposed, but they have dificulty to be directly applied to GWAS caused by excessive search space and intensive computational burden. In this paper, we propose an ant colony optimization (ACO) based algorithm by combining the pheromone updating rule with the heuristic information. We tested power performance of our algorithm by conducting suficient experiments including a wide range of simulated datasets experiments and a real genome-wide dataset experiment. Experimental results demonstrate that our algorithm is time efficient and gain good performance in the term of the power of prediction accuracy. Copyright 2014 ACM.

关键词： Ant colony optimization

来源：评论

学校读者我要写书评

暂无评论

Detecting epistasis by LASSO-penalized-model search algorithm in human Genome-Wide Association Studies

Detecting epistasis by LASSO-penalized-model search algorith...

引用

4th International Conference on Materials Science and Information Technology, MSIT 2014

作者： Zhou, Zhi Hui Liu, Gui Xia Su, Ling Tao Han, Liang Yan, Lun College of Computer Science and Technology Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Jilin University Changchun 130012 China

ISBN: (纸本)9783038351733

Extensive studies have shown that many complex diseases are influenced by interaction of certain genes, while due to the limitations and drawbacks of adopting logistic regression (LR) to detect epistasis in human Genome-Wide Association Studies (GWAS), we propose a new method named LASSO-penalized-model search algorithm (LPMA) by restricting it to a tuning constant and combining it with a penalization of the L1-norm of the complexity parameter, and it is implemented utilizing the idea of multi-step strategy. LASSO penalized regression particularly shows advantageous properties when the number of factors far exceeds the number of samples. We compare the performance of LPMA with its competitors. Through simulated data experiments, LPMA performs better regarding to the identification of epistasis and prediction accuracy. © (2014) Trans Tech Publications, Switzerland.

关键词： Genes

来源：评论

学校读者我要写书评

暂无评论

Density and sliding window-based clustering over evolving data streams

引用

ICIC Express Letters, Part B: Applications 2015年第8期6卷 2275-2283页

作者： Yu, Yanwei Zhao, Jindong Zhang, Yonggang Wen, Changci School of Computer and Control Engineering Yantai University No. 32 Qingquan Road Laishan District Yantai China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Jilin University No. 2699 Qianjin Street Changchun China Jingchang Office of Chinese Airforce No. 11 Shuangyuan Road Shijingshan District Beijing China

Density-based clustering over huge volumes of evolving data streams is critical for many modern applications ranging from network traffic monitoring to moving object management. In this work, we propose an efficient density-based clustering algorithm under sliding window environments to handle evolving data streams. First, we use cell partition optimization to fast determine points’ status instead of expensive completed neighbor query for core points. Second, we cluster new and expired points with cell-cluster structure. Cell-cluster leverages relationships of connected cells around points to complete classification of the points in the cells. Under the effect analysis of new and expired points, we then design an exact clustering framework with cell partition and cell-cluster in streaming windows. Our experimental evaluation, using both real and synthetic streaming data, demonstrates that our algorithm shows effectiveness in capturing the clusters with noise and performs faster than state-of-the art methods, and scales to high volume data streams. © 2015, ICIC International.

关键词： Cells

来源：评论

学校读者我要写书评

暂无评论

An object shrunken algorithm for image classification

引用

Journal of computational Information Systems 2014年第20期10卷 8741-8748页

作者： Pan, Zheng Ouyang, Dantong College of Computer Science and Technology Jilin University ChangChun China Key Laboratory of Computation and Knowledge Engineering Ministry of Education ChangChun China

In this letter, we suggest a novel Object Shrunken (OS) algorithm to handle the image classification task. Unlike the prior art, this letter considers the foreground or the object location in the image for more discriminative image-level representation. The OS algorithm suggests a straightforward procedure to box the object location. It first proposes a Weighted Local Outlier Factor (WLOF) to remove all the interest point outliers, and then positions the object location in terms of the distribution of the rest interest points. We evaluate the proposed algorithm on the well-known dataset Caltech-101. The resulting OS algorithm outperforms the state-of-art approaches in the image classification task.

关键词： Image classification

来源：评论

学校读者我要写书评

暂无评论

Boosting image retrieving result with GPS sensor information

引用

Sensor Letters 2014年第2期12卷 213-217页

作者： Zhao, Hongwei Wang, Zhen Liu, Pingping Gao, Shan Nong, Zhimeng China School of Computer Science and Technology Jilin University Changchun China Key Laboratory of Symbolic Computation and Knowledge Engineering Ministry of Education No. 2699 Qianjin Street Chaoyang District Changchun China

Due to the limited computing resource on mobile devices, it is more difficult to get semantically relevant results from a large dataset in time in a mobile image retrieving system, compared with normal content based image retrieval system, such as in a PC. In order to fix up this problem, images annotated with GPS sensor information have been proposed in this *** an image is acquired by a mobile device, the corresponding location information deriving from GPS coordinates can be obtained by the GPS sensor embedded in the mobile device. For the purpose of using the GPS information, the index database should be built first. In such an indexed database, the whole image database is divided into several small independent databases according to images' location label. Depending on this indexed database and the location label of the query image, image retrieval is carried in a small dataset. To further accelerate the retrieval speed, the hash-based image feature is adopted in this paper. Experimental results show that our proposed method has excellent retrieval performance and low computational complexity. Copyright © 2014 American Scientific Publishers.

关键词： Global positioning system

来源：评论

学校读者我要写书评

暂无评论

A directional algorithm based on shape context for 3D point pattern matching

引用

Journal of computational Information Systems 2015年第7期11卷 2355-2365页

作者： Pang, Kuo Liang, Yanchun Zhou, You Wang, Kangping Wang, Shu Key Laboratory for Symbol Computation and Knowledge Engineering of National Education Ministry College of Computer Science and Technology Jilin University Changchun China Department of Computer Science and Technology Zhuhai College of Jilin University Zhuhai China

Point pattern matching is the basis of image recognition and computer vision. Point pattern matching in three dimensional space with the presence of noise and outlier is an important research focus. In this paper, we present a novel method for point pattern matching base on shape context. Particularly, we aim to find robust feature points to determine the positive direction of the point set for three dimensional rotation. At the same time, we improve the feature extraction and similarity measurement, and then combine with bipartite map matching. Comparative experiments on both synthetic point sets and real data sets demonstrate that the proposed algorithm is effective and robust. ©, 2015, Binary Information Press. All right reserved.

关键词： Image recognition

来源：评论

学校读者我要写书评

暂无评论

Reliability redundancy optimization algorithm based on eagle strategy and PSO

International Journal of Multimedia and Ubiquitous Engineeri...

引用

International Journal of Multimedia and Ubiquitous engineering 2014年第12期9卷 375-382页

作者： Liu, Yubao Qin, Guihe College of Computer Science and Technology Jilin University Changchun China College of Computer Science and Technology Changchun University Changchun China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Jilin University Changchun China

A new algorithm combined Eagle Strategy with PSO is proposed. The new algorithm performs by two phases: First Eagle Strategy is used to do global search;Second PSO algorithm is used to do fast local search around a promising solution. The balance of global search and local search is considered simultaneously. It is not only help to jump out of local optimum but also accelerate local convergence. Experimental results on three benchmark problems illustrate that the presented approach is effective, efficient and accurate for solving reliability redundancy optimization problems. Copyright © 2014 SERSC

关键词： Nonlinear programming

来源：评论

学校读者我要写书评

暂无评论

Comparative Study on Metaheuristic-Based Feature Selection for Cotton Foreign Fibers Recognition

Comparative Study on Metaheuristic-Based Feature Selection f...

引用

The 9th International Conference on Computer and Computing Technologies in Agriculture (CCTA 2015)

作者： Xuehua Zhao Xueyan Liu Daoliang Li Huiling Chen Shuangyin Liu Xinbin Yang Shaobin Zhan Wenyong Zhao School of Digital Media Shenzhen Institute of Information Technology Key Laboratory of Symbolic Computation and Knowledge Engineer (Jilin University) Ministry of Education College of Information and Electrical Engineering China Agricultural University College of Physics and Electronic Information Wenzhou University College of Information Guangdong Ocean University

ISBN: (纸本)9783319483566

The excellent feature set or feature combination of cotton foreign fibers is great significant to improve the performance of machine-vision-based recognition system of cotton foreign fibers. To find the excellent feature sets offoreign fibers, in this paper presents three metaheuristic-based feature selection approaches for cotton foreign fibers recognition, which are particle swarm optimization, ant colony optimization and genetic algorithm, respectively. The k-nearest neighbor classifier and support vector machine classifier with k-fold cross validation are used to evaluate the quality offeature subset and identify the cotton foreign fibers. The results show that the metaheuristic-based feature selection methods can efficiently find the optimal feature sets consisting of a few features. It is highly significant to improve the performance of recognition system for cotton foreign fibers.

关键词： Metaheuristic Feature selection Foreign fibers Recognition system Feature extraction recognition system Fiber Fibrous materials Cotton Support Vector Network

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：