检索结果-南通市图书馆

discriminative learning with Scale Decomposition for Person Detection

Wuhan University Journal of Natural Sciences 2020年第4期25卷 337-342页

作者： WANG Xiao CHEN Jun LIANG Chao HU Ruimin National Engineering Research Center for Multimedia Software Wuhan UniversityWuhan 430072HubeiChina

Person detection,which can locate the person regions in the image,continues to be a hot research topic in both computer vision and signal processing ***,detecting person at small scale remains a challenging problem due to the lack of discriminative details in the typical image at small *** this paper,we propose a decomposition mapping method which contains two subnets:encoder subnet and decoder *** subnet can exploit decomposition transformation for person regions from big scale to small *** subnet reverses the process of the encoder *** add deconvolution network to the decoder subnet to make up for the lost information and a discriminative mapping has been restructured to transform the person regions from the small scale to the big ***,person-regions and background-regions can then be separated according to their decomposition positions in the new scale *** proposed approach is evaluated on two challenging person datasets:Caltech dataset and the KITTI *** with SAF R-CNN,the miss rate has been optimized by 3.96%on Caltech person dataset and the mean average precision has been optimized by 1.76%on KITTI person dataset.

关键词： discriminative learning scale decomposition person detection

来源：

维普期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

A new constrained maximum margin approach to discriminative learning of Bayesian classifiers

引用

Frontiers of Information Technology & Electronic Engineering 2018年第5期19卷 639-650页

作者： Ke GUO Xia-bi LIU Lun-hao GUO Zong-jie LI Zeng-min GENG Beijing Laboratory of Intelligent Information Technology School of Computer ScienceBeijing Institute of Technology Computer Information Center Beijing Institute of Fashion Technology

We propose a novel discriminative learning approach for Bayesian pattern classification, called ＇constrained maximum margin （CMM）＇. We define the margin between two classes as the difference between the minimum decision value for positive samples and the maximum decision value for negative samples. The learning problem is to maximize the margin under the con- straint that each training pattern is classified correctly. This nonlinear programming problem is solved using the sequential un- constrained minimization technique. We applied the proposed CMM approach to learn Bayesian classifiers based on Gaussian mixture models, and conducted the experiments on 10 UCI datasets. The performance of our approach was compared with those of the expectation-maximization algorithm, the support vector machine, and other state-of-the-art approaches. The experimental results demonstrated the effectiveness of our approach.

关键词： discriminative learning Statistical modeling Bayesian pattern classifiers Gaussian mixture models UCI datasets

来源：

维普期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

Robust Speech Recognition Method Based on discriminative Environment Feature Extraction

引用

Journal of Computer Science & Technology 2001年第5期16卷 458-464页

作者：韩纪庆高文 DepartmentofComputerScienceandEngineering HarbinInstituteofTechnologyHarbin150001P.R.China InstituteofComputingTechnology TheChineseAcademyofSciencesBeijing100080P.R.China

It is an effective approach to learn the influence of environmental parameters, such as additive noise and channel distortions, from training data for robust speech recognition. Most of the previous methods are based on maximum likelihood estimation criterion. However, these methods do not lead to a minimum error rate result. In this paper, a novel discrimina-tive learning method of environmental parameters, which is based on Minimum Classification Error (MCE) criterion, is proposed. In the method, a simple classifier and the Generalized Probabilistic Descent (GPD) algorithm are adopted to iteratively learn the environmental pa-rameters. Consequently, the clean speech features are estimated from the noisy speech features with the estimated environmental parameters, and then the estimations of clean speech features are utilized in the back-end HMM classifier. Experiments show that the best error rate reduction of 32.1% is obtained, tested on a task of 18 isolated confusion Korean words, relative to a conventional HMM system.

关键词： robust speech recognition minimum classification error environmental parameter discriminative learning

来源：

维普期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

Multi-View Non-negative Matrix Factorization Discriminant learning via Cross Entropy Loss

Multi-View Non-negative Matrix Factorization Discriminant Le...

引用

第32届中国控制与决策会议

作者： Jian-Wei Liu Yuan-Fang Wang Run-Kun Lu Xiong-Lin Luo Department of Automation China University of PetroleumBeijing Campus(CUP)

Multi-view learning accomplishes the task objectives of classification by leveraging the relationships between different views of the same *** existing methods usually focus on consistency and complementarity between multiple *** not all of this information is useful for classification ***,it is the specific discriminating information that plays an important *** Zhang et al explores the discriminative and non-discriminative information existing in common and view-specific parts among different views via joint non-negative matrix *** this paper,we improve this algorithm on this basis by using the cross entropy loss function to constrain the objective function *** last,we implement better classification effect than original on the same data sets and show its superiority over many state-of-the-art algorithms.

关键词： Multi-View learning Matrix factorization Cross entropy loss discriminative learning Classification

来源： cnki会议评论

在线全文

cnki会议

学校读者我要写书评

暂无评论

Linguistically-enriched Information Retrieval

Linguistically-enriched Information Retrieval

引用

第四届全国信息检索与内容安全学术会议

作者： Haoliang QI~(1,2+) Muyun YANG~1 Jianfeng GAO~3 Sheng LI~1 Guohua LEI~2 ~1 School of Computer Science and Technology,Harbin Institute of Technology,150001,Harbin,China, ~2 Department of Computer Science and Technology,Heilongiiang Institute of Technology,150050,Harbin,China ~3 Microsoft Corporation,Redmond,WA 98052,U.S.A.

There are two problems in linguistically-enriched Information Retrieval(IR),one is how to get the linguistically decencies,the other is how to integrate linguistically features in an IR *** paper presents the reasons why using shallow parsing instead of full parsing to get linguistically *** then scores of linguistically feature are assigned by Hidden Markov Model(HMM),which is widely used in language modeling approaches to *** average precision training is used to training the free parameters of the linear discriminative model,which optimizes the average precision(AP)directly using an iterative *** effectiveness of our approach has been evaluated on the task of ad hoe retrieval using 5 English and Chinese TREC test *** show that the presented method significantly outperforms the state-of-the-art language modeling approaches and the classical probabilistic retrieval model.

关键词： Information Retrieval Linguistically Features Chunking Generative Feature Extraction discriminative learning

来源： cnki会议评论

在线全文

cnki会议

学校读者我要写书评

暂无评论

欢迎您,

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

在线全文

在线全文

在线全文

在线全文

在线全文

请选择保存的检索档案：

请选择收藏分类：

通借通还

欢迎您,

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

在线全文

在线全文

在线全文

在线全文

在线全文

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：