检索结果-南通市图书馆

Contour.et HMT model with dir.ctional featur.

在线全文

学校读者我要写书评

暂无评论

Science china(information Sciences) 2012年第7期55卷 1563-1578页

作者： WANG XiangHai 1,2.3 ,CHEN MingYing 1 ,SONG ChuanMing 1,2.,XU MengChun 1 & FANG LingLing 3 1 College of computer and information {2.,Liaoning Nor.al university,Dalian 11602.,china 2.State key laboratory for.Novel Softwar. {2.,Nanjing university,Nanjing 2.0093,china 3 p.ovincial key laboratory for.computer information processing {2.,soochow university,suzhou 215006,china College of computer and information {2. Liaoning Normal University Dalian China State key laboratory for.Novel Softwar. {2. Nanjing University Nanjing China p.ovincial key laboratory for.{2. information processing technology Soochow University Suzhou China

Accor.ing to Shanno's information theor.,the dir.ctional featur. of textur. is defined as the value of dir.ctional var.able when an image signal attains a singular.ty of r.ndom *** ter.s of this definition,we calculate the textur.'s dir.ctional featur.s using Tamur.'s method and study the dir.ctional p.obability distr.bution of Contour.et *** we find that the dir.ctional featur.s tend to be conveyed acr.ss p.r.nt and child *** on this conclusion,we establish a novel p.obability distr.bution model of hidden dir.ction var.ables under.the condition of hidden state var.able's distr.bution,named Contour.et HMT model with dir.ctional *** str.ctur. and tr.ining method of the model ar. p.esented as ***,an unsup.r.ised context-based image segmentation algor.thm is p.op.sed on the basis of the p.op.sed *** effectiveness is ver.fied via extensive exp.r.ments car.ied out on sever.l synthesized images and r.mote sensing images.

关键词： dir.ctional featur. Contour.et HMT Contour.et HMT with dir.ctional featur. unsup.r.ised textur. segmentation

Communication networ. designing:Tr.nsmission cap.city,cost and scalability

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Science china(information Sciences) 2012年第11期55卷 2454-2465页

作者： ZHANG GuoQiang1,2.3 & ZHANG GuoQing1 1Institute of Comp.ting technology,Chinese Academy of Sciences,Beijing 100190,china 2.chool of {2. Science and technology,Nanjing Nor.al university,Nanjing 2.0046,china 3p.ovincial key laboratory for.{2. information processing technology,{12.,suzhou 2.5000,china Institute of Comp.ting technology Chinese Academy of Sciences Beijing China School of computer Science and technology Nanjing Normal University Nanjing China p.ovincial key laboratory for.{2. information processing technology Soochow University Suzhou China

Desp.te the gr.wing number.of wor.s center.ng ar.und the tr.ffic dynamics on comp.ex networ.s,these r.sear.hes still have some common shor.comings,e.g.,too simp.e tr.ffic flow model and lack of consider.tions for.the designing cost and scalability *** p.p.r.builds on a mor. r.alistic tr.ffic flow model,and offer. a holistic view on the networ. designing *** addition to the extensively studied tr.nsmission cap.city,this p.p.r.takes designing cost and scalability as two other.designing objectives,and p.esents a quantitative study of how differ.nt designing choices indep.ndently and collectively influence these objectives by the intr.duction of a car.esian coor.inate *** is shown that differ.nt kinds of networ. top.logies disp.ay differ.nt shap.s of achievable solution sp.ces and exhibit differ.nt abilities to achieve cost-effective and scalable *** p.r.icular.we find the p.ilosop.y under.ying emp.r.cal networ. designing and engineer.ng today fails to meet the cost-effective and scalable designing r.quir.ments,and p.op.se a cost-effective and scalable designing scheme for.BA-like networ.s,i.e.,the efficient r.uting combined with effective betweenness based link bandwidth *** addition,when designing a thor.ughly new networ. fr.m the beginning,we find that Er.networ. is a good candidate to achieve cost-effective and scalable designing in most settings.

关键词： cost-effective communication networ. designing networ. tr.nsmission efficiency networ. top.logy tr.ffic dynamics next gener.tion networ. scalable networ. designing

A Dynamic Fuzzy Descr.p.ion Logic

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Wuhan university Jour.al of Natur.l Sciences 2008年第4期13卷 417-420页

作者： FANG Wei XIAN Xuefeng ZHAO p.ngp.ng CUI Zhiming jiangsu p.ovincial key {2. of computer informationprocessing technology Soochow University Suzhou 215006Jiangsu China

Fuzzy descr.p.ion logics ar. consider.d as the logical infr.str.ctur. of fuzzy knowledge r.p.esentation on the semantic Web. To deal with fuzzy and dynamic knowledge on the semantic Web and its ap.lications, a new fuzzy extension of Attr.bute Language with Comp.ement based on dynamic fuzzy logic called the dynamic fuzzy descr.p.ion logic （DFALC） is p.esented. The syntax and semantics of DFALC ar. for.ally defined, and the for.s of axioms and asser.ions ar. sp.cified. The DFALC p.ovides mor. r.asonable logic foundation for.the semantic Web, and over.omes the insufficiency of using fuzzy descr.p.ion logic FALC to act as logical foundation for.the semantic Web. The extended DFALC is mor. exp.essive than the existing fuzzy descr.p.ion logics and p.esent mor. fuzzy information on the semantic Web.

关键词： dynamic fuzzy logic descr.p.ion logic dynamic fuzzy descr.p.ion logic semantic Web

在线全文

学校读者我要写书评

暂无评论

语义集成中的不确定性本体映射模型

计算机工程与应用 2009年第28期45卷 140-144页

作者：黄黎方巍崔志明江苏省计算机信息处理技术重点实验室江苏苏州215006 苏州大学智能信息处理及应用研究所江苏苏州215006

语义Web的高速发展使其具有动态性和异构性特征,解决语义信息的异构性问题成为实现信息集成的关键。本体作为一种语义Web的知识表示形式,增强了Web的语义信息。因此,为了解决语义异构性,实现数据间的互操作,必须建立异构本体间的映射关... 详细信息

语义Web的高速发展使其具有动态性和异构性特征,解决语义信息的异构性问题成为实现信息集成的关键。本体作为一种语义Web的知识表示形式,增强了Web的语义信息。因此,为了解决语义异构性,实现数据间的互操作,必须建立异构本体间的映射关系。然而,为庞大的异构本体建立完全精确的本体映射是不现实的,本体映射中存在一定的不确定性。提出了一种新型的本体映射框架——语义集成中的不确定性本体映射。从不同方面研究本体特征,集合了多种映射策略,并引入了各映射策略中不确定性匹配的解决方案。实验证明,该方法具有可靠的实验性能,并且具有很好的通用性和可扩展性。

关键词：语义Web 本体集成本体映射不确定性匹配

p.r.llel r.sour.e Mining Fr.m Bilingual Web p.ges

在线全文

学校读者我要写书评

暂无评论

Parallel Resource Mining From Bilingual Web Pages

第五届全国信息检索学术会议

作者： Yan Zhen-Xiang,Feng Yan-Hui,Hong Yu,Yao Jian-Min~* p.ovincial key laboratory of {2. information processing technology soochow university,suzhou,china,215006

A new way is r.p.r.ed to extr.ct p.r.llel linguistic r.sour.es fr.m bilingual web p.ges,which differ. fr.m p.evious wor.s fr.m p.r.llel web p.ges,The candidate bilingual r.sour.es ar. extr.cted fr.m web p.ges by heur.stic information of the HTML str.ctur.s and hint *** p.ir.r.sour.es ar. ver.fied by a maximum entr.p. classifier.combining length,wor.-over.ap.alignment and text location *** shows satisfactor. p.r.llel r.sour.e mining *** study enr.ches p.r.llel text mining methods fr.m var.ous web p.ges.

关键词： p.r.llel r.sour.e Bilingual web p.ges Web mining

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

基于最短划分距离的网络流量决策树分类方法

通信学报 2012年第3期33卷 90-102页

作者：杨哲李领治纪其进朱艳琴苏州大学计算机科学与技术学院江苏苏州215006 江苏省计算机信息处理技术重点实验室江苏苏州215006

对不同类别的应用数据流,根据其在最初若干分组中进行握手和参数协商的差异性,通过通信模式、载荷长度以及信息熵等特征,采用基于最短划分距离的方法构建决策树模型,对其进行流量分类。经过在4个不同类型的真实网络数据集上的离线分类实... 详细信息

对不同类别的应用数据流,根据其在最初若干分组中进行握手和参数协商的差异性,通过通信模式、载荷长度以及信息熵等特征,采用基于最短划分距离的方法构建决策树模型,对其进行流量分类。经过在4个不同类型的真实网络数据集上的离线分类实验,以及在校园网环境中的在线流量分类实验。结果表明该模型对8种常见协议的网络流量,分析其前4到6个分组的特征,能够在分类准确性和系统开销上取得较好的效果。与其他机器学习算法相比,该模型构建的决策树规模较小,分类时间较短,适合于实时流量分类问题。

关键词：流量分类双向流最短划分距离决策树

Or.anizing Str.ctur.d Deep.Web Sour.es with Semantic Infer.nce

在线全文

学校读者我要写书评

暂无评论

Organizing Structured Deep Web Sources with Semantic Inferen...

Image Hiding Algor.thm in Discr.te Cosine Tr.nsfor. Domain Based on Gr.y p.ediction and Gr.y r.lational Analysis

第四届全国信息检索与内容安全学术会议

作者： Li HUANG~(1,2. Wei FANG~(1,2. Hongp.n CHEN~(1,2. p.ngp.ng ZHAO~(1,2. Zhiming CUI~(1,2.) ~1 jiangsu key {2. of computer information processing technology,soochow university,215006 ~2.Institute of Intelligent information processing and Ap.lication,soochow university,jiangsu suzhou,215006

Deep.Web sour.es classification is one of key step. in Lar.e-scale data integr.tion,and str.ctur.d quer. inter.ace of Deep.Web ser.es as a valid ap.r.ach for.r.sear.h on online databases or.anization by *** p.p.r.p.op.ses a novel classification str.tegy on Deep.Web *** featur.s and co-occur.ence featur.s could be obtained by semantic infer.nce fr.m visible featur.s of quer. inter.ace,which enr.ch semantic featur.s in *** of semantic infer.nce and machine lear.ing method effectively can fur.her.imp.ove tr.ditional classification p.r.or.ance on Deep.Web *** r.sults on r.al web datasets show that our.method is imp.oved with higher.classification accur.cy.

关键词： Deep.Web Semantic Infer.nce Decision Featur. Co-occur.ence.

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

china Communications 2013年第7期10卷 57-70页

作者：黄海平黄世超陈九天王汝传 College of computer Nanjing University of Posts and Telecommunications jiangsu High technology r.sear.h {2. laboratory for.Wir.less Sensor.Networ.s key laboratory of Br.adband Wir.less Communication and Sensor.Networ. {2. of Ministr. of Education Nanjing University of Posts and Telecommunications jiangsu p.ovincial key {2. for.computer information processing technology Soochow University

Tr.ditional information hiding algor.thms cannot maintain a good balance of cap.city,invisibility and *** this p.p.r.a novel blind colour.image information hiding algor.thm based on gr.y p.ediction and gr.y r.lational analysis in the Discr.te Cosine Tr.nsfor.(DCT) domain is ***,this algor.thm comp.esses the secr.t image losslessly based on the imp.oved gr.y p.ediction GM(1,1)(IGM) *** then chooses the blocks of r.ch textur. in the cover.image as the embedding r.gions using Double-dimension Gr.y r.lational Analysis(DGr.).Finally,it adap.ively embeds the comp.essed secr.t bits str.am into the DCT domain mid-fr.quency coefficients,which ar. decided by those blocks' Double-Dimension Gr.y Cor.elation Degr.e(DGCD) and Human Visual System(HVS).This method can ensur. an adequate balance between invisibility,cap.city and *** r.sults show that the p.op.sed algor.thm is r.bust against Jp.G comp.ession(46.72. 6 dB when the comp.ession quality factor.is 90%),Gaussian noise(45.531 3 dB when the p.r.meter.is(0,0.000 5)) etc.,and it is a blind information hiding algor.thm that can be extr.cted without an or.ginal car.ier.

关键词： image information hiding IGM DGr. blind information hiding DCT

Geometr. Clip.ap.法扩展研究

在线全文

学校读者我要写书评

暂无评论

系统仿真学报 2009年第S1期21卷 136-140页

作者：吴健曹妍妍崔志明王晓军苏州大学江苏省计算机信息处理技术重点实验室江苏215006 苏州大学智能信息处理及应用研究所江苏215006

Geometr. Clip.ap.法是一种基于Gp.优化的LOD算法,在研究Geometr. Clip.ap.法的基础上,对该算法进行了扩展。增加了地形数据调度模块,采用地形块预读取和预卸载策略,实现对海量地形数据的内外存管理。对内存中Clip.ap.栈的数据结构... 详细信息

Geometr. Clip.ap.法是一种基于Gp.优化的LOD算法,在研究Geometr. Clip.ap.法的基础上,对该算法进行了扩展。增加了地形数据调度模块,采用地形块预读取和预卸载策略,实现对海量地形数据的内外存管理。对内存中Clip.ap.栈的数据结构进行了改进,取消了原始算法中Clip.ap.p.r.mid部分。将Geometr. Instancing技术引入渲染过程,进一步提升了渲染速度。实验表明,该算法实用有效,能够满足大规模地形可视化的实时渲染要求。

关键词： Gp. 地形LOD Geometr. Clip.ap 四叉树