简介：Thissurveyreviewstherecentdevelopmentofgradientdomainmeshdeformationmethod.Differenttootherdeformationmethods,thegradientdomaindeformationmethodisasurface-based,variationaloptimizationmethod.Itdirectlyencodesthegeometricdetailsindifferentialcoordinates,whicharealsocalledLaplaciancoordinatesinliterature.BypreservingtheLaplaciancoordinates,themeshdetailscanbewellpreservedduringdeformation.DuetothelocalityoftheLaplaciancoordinates,thevariationaloptimizationproblemcanbecastedintoasparselinearsystem.Fastsparselinearsolvercanbeadoptedtogeneratedeformationresultinteractively,oreveninreal-time.Thenonlinearnatureofgradientdomainmeshdeformationleadstothedevelopmentoftwocategoriesofdeformationmethods:linearizationmethodsandnonlinearoptimizationmethods.Basically,thelinearizationmethodsonlyneedtosolvethelinearleast-squaressystemonce.Theyarefast,easytounderstandandcontrol,whilethedeformationresultmightbesuboptimal.Nonlinearoptimizationmethodscanreachoptimalsolutionofdeformationenergyfunctionbyiterativeupdating.Sincethecomputationofnonlinearmethodsisexpensive,reduceddeformablemodelsshouldbeadoptedtoachieveinteractiveperformance.Thenonlinearoptimizationmethodsavoidtheuserburdentoinputtransformationatdeformationhandles,andtheycanbeextendedtoincorporatevariousnonlinearconstraints,likevolumeconstraint,skeletonconstraint,andsoon.Wereviewrepresentativemethodsandrelatedapproachesofeachcategorycomparativelyandhopetohelptheuserunderstandthemotivationbehindthealgorithms.Finally,wediscusstherelationbetweenphysicalsimulationandgradientdomainmeshdeformationtorevealwhyitcanachievephysicallyplausibledeformationresult.

简介：情绪分类上的State-of-the-arts研究典型地是域依赖者并且限制域。在这份报纸，我们试图减少领域相关性并且由建议一个有效多域情绪分类算法同时改进全面性能。我们的方法采用多重分类器联合的途径。在这条途径，我们首先与领域独立训练单个领域分类器特定的数据，然后为最后的决定联合分类器。我们的实验比两个挑选领域分类途径的证明这条途径更好表现(个别地使用训练数据)并且混合领域分类途径(都简单地联合训练数据)。特别地，有加权的和统治的分类器联合在单个领域分类上获得27.6%的平均错误减小。

简介：Asteadyincreaseinconsumerdemands,andsevereconstraintsfrombothasomewhatdamagedenvironmentandnewlyinstalledgovernmentpolicies,requiretoday'sproductdesignanddevelopmenttobefasterandmoreefficientthaneverbefore,yetutilizingevenfewerresources.Newholisticapproaches,suchastotalproductlifecyclemodelingwhichembracesallaspectsofaproduct'slifecycle,arecurrentattemptstosolvetheseproblems.Withinthefieldofproductdesignandmodeling,featuretechnologyhasprovedtobeoneverypromisingsolutioncomponent.Owingtothetremendousincreaseininformationtechnology,totransferfromlowleveldataprocessingtowardsknowledgemodelingandinformationprocessingisabouttobringachangeinalmosteverycomputerizedapplication.Fromthisviewpoint,currentproblemsofbothfeatureframeworksandfeaturesystemsareanalyzedinrespecttostaticanddynamicconsistencybreakdowns.Theanalysisrangesfromearlystagesofdesigning(feature)conceptstofinalsystemimplementationandapplication.Forthefirsttime,anintegratedviewisgivenonapproaches,solutionsandpracticalexperience,withfeatureconceptsandstructures,providingbothafeatureframeworkanditsimplementationwithsufficientsystemarchitectureandcomputationalpowertomasterafairnumberofknownconsistencybreakdowns,whileprovidingforrobustcontextsforfeaturesemanticsandintegratedmodels.Withintoday'sheavyuseofinformationtechnologythesearepre-requisitesifthefullpotentialoffeaturetechnologyistobesuccessfullytranslatedintopractice.

简介：中国microblogs的意见目标抽取在意见采矿起一个重要作用。最近在这个区域有重要进步，特别方法基于有条件的随机的地(CRF)。然而，这个方法仅仅考虑词典相关的特征并且不挖掘暗示了语法、语义的知识。我们建议与一些句法、语义的特征一起合并领域词典的一条新奇途径。途径通过基于词嵌入通过Partof讲话，相关性结构，短语结构，语义角色和语义类似探索语法、语义的信息的一个新奇方法获得领域词典。然后我们把领域词典与意见目标为意见目标抽取与一些特征一起从CRF提取了相结合。COAE2014数据集上的试验性的结果在意见目标抽取的任务与另外的著名方法相比显示出途径的outperformance。