简介：一位英国人MrsHarris到中国工作，住到了王先生的隔壁，和王先生成了邻居。一开始他们相处得很融洽，但是好景不长，麻烦事就多了。MrsHarris给王先生写了一封信，我们来看看发生了什么事情。

简介：TheTiantaNeighborhoodCommunityCenterinNorthChina'sTianjinMunicipalityhascreatednewactivities—NeighborFestivalaspartoftheeffortstorallythepeoplelivinginthecommunityandbuildaharmoniousneighborhood.Since1999,thefestivalhasbecomeanannualevent.Activitiesincludelectures,forums,theatricalperformances,funsports,talentshows,povertyreliefandmutualaid.Alltheactivitiesarecenteredroundonetheme:goodneighbors,unityandmutualaid,saidXianShuwen,headoftheTiantaNeighborhoodCommunityOffice.Thepurposeistomakeallpeopleliveinthecommunityharmoniouslytogether.In

简介：Itiswell-knownthatinordertobuildastrongensemble,thecomponentlearnersshouldbewithhighdiversityaswellashighaccuracy.Ifperturbingthetrainingsetcancausesignificantchangesinthecomponentlearnersconstructed,thenBaggingcaneffectivelyimproveaccuracy.However,forstablelearnerssuchasnearestneighborclassifiers,perturbingthetrainingsetcanhardlyproducediversecomponentlearners,thereforeBaggingdoesnotworkwell.ThispaperadaptsBaggingtonearestneighborclassifiersthroughinjectingrandomnesstodistancemetrics.Inconstructingthecomponentlearners,boththetrainingsetandthedistancemetricemployedforidentifyingtheneighborsareperturbed.AlargescaleempiricalstudyreportedinthispapershowsthattheproposedBagInRandalgorithmcaneffectivelyimprovetheaccuracyofnearestneighborclassifiers.

简介：

简介：TheproblemofcontinuouslymonitoringmultipleK-nearestneighbor(K-NN)querieswithdynamicobjectandquerydatasetisvaluableformanylocation-basedapplications.Apracticalmethodistopartitionthedataspaceintogridcells,withbothobjectandquerytablebeingindexedbythisgridstructure,whilesolvingtheproblembyperiodicallyjoiningcellsofobjectswithquerieshavingtheirinfluenceregionsintersectingthecells.Intheworstcase,allcellsofobjectswillbeaccessedonce.ObjectandquerycachestrategiesareproposedtofurtherreducetheI/Ocost.Withobjectcachestrategy,queriesremainingstaticincurrentprocessingcycleseldomneedI/Ocost,theycanbereturnedquickly.ThemainI/Ocostcomesfrommovingqueries,thequerycachestrategyisusedtorestricttheirsearch-regions,whichusescurrentresultsofqueriesinthemainmemorybuffer.Thequeriescansharenotonlytheaccessingofobjectpages,butalsotheirinfluenceregions.TheoreticalanalysisoftheexpectedI/Ocostispresented,withtheI/Ocostbeingabout40%thatoftheSEA-CNNmethodintheexperimentresults.

简介：第四级的代码的特征被分析。区分三角方向的规则被给出。邻居由从回到前面分解第四级的代码发现的一个算法在这篇论文被介绍。在这个算法和Bartholdi之间的时间复杂性的对比的分析是算法被接近。结果说明一般水准消费了这个算法的时间，这是大约23.66%Bartholdi的算法。

简介：Letbethecollectionofm-timescontinuouslydifferentiableprobabilitydensitiesfonRdsuchthat丨Daf（x1）-Daf（x2）丨≤M‖x1-x2‖βforx1,x2∈Rd,[a]=m,whereDadenotesthedifferentialoperatordefinedbyDa=（[a]）/（x1a…xdad）.UnderratherweakconditionsonK（x）,thenecessaryandsufficientconditionsforsup丨n（x）-f（x）丨=0((（logn/n）λ/（d+3λ）,λ=m+β,f∈arethat∫xaK（xi）dx=0for0<[a]≤m.Finallytheconvergencorateatapointisgiven.

简介：Inthispaper,anewapproachispresentedtofindthereferencesetforthenearestneighborclassifier.Theoptimalreferenceset,whichhasminimumsamplesizeandsatisfiesacertainerrorratethreshold,isobtainedthroughaTabusearchalgorithm.Whentheerrorratethresholdissettozero,thealgorithmobtainsanearminimalconsistentsubsetofagiventrainingset.Whilethethresholdissettoasmallappropriatevalue,theobtainedreferencesetmaycompensatethebiasofthenearestneighborestimate.AnaspirationcriterionforTabusearchisintroduced,whichaimstopreventthesearchprocessfromtheinefficientwanderingbetweenthefeasibleandinfeasibleregionsinthesearchspaceandspeeduptheconvergence.Experimentalresultsbasedonanumberoftypicaldatasetsarepresentedandanalyzedtoillustratethebenefitsoftheproposedmethod.Comparedtoconventionalmethods,suchasCNNandDasarathy'salgorithm,thesizeofthereducedreferencesetsismuchsmaller,andthenearestneighborclassificationperformanceisbetter,especiallywhentheerrorratethresholdsaresettoappropriatenonzerovalues.TheexperimentalresultsalsoillustratethattheMCS(minimalconsistentset)ofDasarathy'salgorithmisnotminimal,anditscandidateconsistentsetisnotalwaysensuredtoreducemonotonically.Acounterexampleisalsogiventoconfirmthisclaim.

简介：LetXbead-dimensionalrandomvectorwithunknowndensityfunctionf（z）=f（z1,···,zd）,andletfnbetehnearestneighborestimatoroffproposedbyLoftsgaardenandQuesenberry（1965）.Inthispaper,weestablishedthelawoftheiteratedlogarithmoffnforgeneralcaseofd≥1,whichgivestheexactpointwisestrongconvergencerateoffn.

简介：Weinvestigatetheconsistencyandasymptoticnormalityofnearest-neighbordensityestimatorofasampledataprocessbasedonα-mixingassumption.Weextendthecorrespondentresultunderindependentidenticalcases.

简介：Wepresentsomenewmethodsforparameterizingthetrianglemeshsurface(TMS)whichresultfromtheMarchingCubes(MC)algorithm.Themethodsapplytosurfacesofgenuszeroandtheparameterdomainisaunitsphere.WetakeadvantageofsomespecialpropertiesoftheTMSresultingfromtheMCalgorithmtoobtainsimple,computationalefficientrepresentationsofthenearestneighborcoordinatesandutilizethesecoordinatesinthecharacterizationoftheparameterizationbymeansofsystemsofequationswhicharesolvediteratively.Examplesandcomparisonsarepresented.

简介：

简介：Aproper[h]-totalcoloringcofagraphGisapropertotalcoloringcofGusingcolorsoftheset[h]={1,2,...,h}.Letw(u)denotethesumofthecoloronavertexuandcolorsonalltheedgesincidenttou.Foreachedgeuv∈E(G),ifw(u)≠w(v),thenwesaythecoloringcdistinguishesadjacentverticesbysumandcallitaneighborsumdistinguishing[h]-totalcoloringofG.Bytndi(G),wedenotethesmallestvaluehinsuchacoloringofG.Inthispaper,weobtainthatGisagraphwithatleasttwovertices,ifmad(G)<3,thentndi∑(G)≤k+2wherek=max{Δ(G),5}.Itpartiallycon?rmstheconjectureproposedbyPil′sniakandWozniak.

简介：Undertheframeworkofsupportvectormachines,thispaperproposesanewkernelmethodbasedonneighborbandsmutualinformationforhyperspectraldatumclassification.Thisalgorithmassignsweightstodifferentbandsinthekernelfunctionaccordingtotheamountofusefulinformationthattheycontain,whichmakesthebandwithmoreusefulinforma-tionplaymoreimportantroleintheclassification.Ourresearchhasshownthatthebandwithgreatermutualinformationbetweenneighborbandscontainsmoreusefulinformation,andhenceweusethemutualinformationofeachbandanditsneighborbandsastheweightsoftheproposedkernelmethod.Theexperimentalresultsshowthatforthesupportvectormachinesbasedonpolynomialandradialbasisfunction,afterintroducingtheproposedkernelfunction,theaverageaccuracyisincreasedmorethan1.2%withoutusinganyreferencemaporincreasingmuchmorecomputationaltime.

简介：all-k-nearest-neighbor(AkNN)询问为每个询问对象发现k最近的邻居。这个问题在许多区域自然地产生，例如GIS(地理信息系统)，多媒体检索，和recommender系统。到各种各样的数据类型和灵活距离度量标准在真实应用程序包含了的支持，也就是，我们在公制的空格学习AkNN检索公制的AkNN(MAkNN)搜索。认为询问上的内在的索引设定，目标集合不能存在，它在许多情形是自然的。例如，询问集合和对象集合能是另外的询问的结果，并且这样，内在的索引不能预先被造。到没有任何内在的索引的数据集上的支持MAkNN搜索，我们建议一个有效基于磁盘的算法，作为基于分区的MAkNN算法(PMA)称为，它跟随一个分区搜索框架并且为加速搜索采用一系列修剪的规则。另外，我们扩大我们的技术处理MAkNN询问的有趣的变体，即，公制的self-AkNN(MSAkNN)搜索，在询问集合与对象相同的地方，设定。用真实、合成的数据集的广泛的实验表明我们的修剪的规则的有效性和建议算法的效率，与最先进的MAkNN和MSAkNN算法相比。

简介：许多真实世界的问题是动态的，要求能连续地追踪随着时间的过去改变optima(最佳)的优化算法。这份报纸用近邻居的效果的观点建议一个改进微分进化算法决定一位个人邻居，为追踪在动态环境的多重optima。用近邻居的效果的新变化策略也被介绍。它由在它的邻居利用存储记忆点，并且利用生产由的微分向量创造个人nearneighbor优异并且near-neighbor-inferior。从生物免疫系统拿灵感，一个免疫系统基于计划为很快检测并且对环境变化作出回应被介绍。另外，一个差别相关的multidirectional扩大计划被介绍从不同尺寸集成珍贵信息为有效地并且很快在搜索空间发现有希望的最佳。动态情形的实验由典型动态测试instancemoving山峰问题创造了，证明了近邻居并且基于的免疫系统微分进化算法(NIDE)在处理动态优化功能是有效的。

简介：G-proteincoupledreceptors(GPCRs)areaclassofseven-helixtransmembraneproteinsthathavebeenusedinbioinformaticsasthetargetstofacilitatedrugdiscoveryforhumandiseases.AlthoughthousandsofGPCRsequenceshavebeencollected,theligandspecificityofmanyGPCRsisstillunknownandonlyonecrystalstructureoftherhodopsin-likefamilyhasbeensolved.Therefore,identifyingGPCRtypesonlyfromsequencedatahasbecomeanimportantresearchissue.Inthisstudy,anoveltechniqueforidentifyingGPCRtypesbasedontheweightedLevenshteindistancebetweentworeceptorsequencesandthenearestneighbormethod(NNM)isintroduced,whichcandealwithreceptorsequenceswithdifferentlengthsdirectly.Inourexperimentsforclassifyingfourclasses(acetylcholine,adrenoceptor,dopamine,andserotonin)oftherhodopsin-likefamilyofGPCRs,theerrorratesfromtheleave-one-outprocedureandtheleave-half-outprocedurewere0.62%and1.24%,respectively.Theseresultsarepriortothoseofthecovariantdiscriminantalgorithm,thesupportvectormachinemethod,andtheNNMwithEuclideandistance.

简介：Thequantumsolitarywavesolutionsinaone-dimensionalferromagneticchainisinvestigatedbyusingtheHartree-Fockapproachandthemultiple-scalemethod.Itisshownthatquantumsolitarywavesolutionscanexistinaferromagneticsystemwithnearest-andnext-nearest-neighborexchangeinteraction,andatthecertainvalueofthefirstBrillouinzone,thesolitarywavesolutionoftheHartreewavefunctionbecomestheintrinsiclocalizedmode.

简介：Inthispaper,anewlatticehydrodynamicmodelisproposedbyincorporatingthedriveranticipationefectofnext-nearest-neighborsite.Thelinearstabilityanalysisandnonlinearanalysisshowthatthedriveranticipationefectofnext-nearest-neighborsitecanenlargethestableareaoftrafcflow.Thespacecanbedividedintothreeregions:stable,metastable,andunstable.Numericalsimulationfurtherilluminatesthatthedriveranticipationefectofthenext-nearest-neighborsitecanstabilizetrafcflowinourmodifiedlatticemodel,whichisconsistentwiththeanalyticalresults.

简介：IOntheoccasionofthe50thanniversaryofthefoundingofthePeople’SRe-publicofChinaandoftheestablishmentofdiplomatictiesbetweenthePRCandRussia,bothcountrieshavepledgedtotakethisopportunitytofurtherstrengthenanddevelopbilateralrelationsinthecoming21stcentury.Lookingbackatthe50-yearrelationshipbetweenChinaandRussia,itisnotdifficulttoseethatitunderwentnotonlyclosecooperation,butalsoseverecon-