简介:稀疏表示是近年来新兴的一种数据表示方法,是对人类大脑皮层编码机制的模拟。稀疏表示以其良好的鲁棒性、抗干扰能力、可解释性和判别性等优势,广泛应用于模式识别领域。基于稀疏表示的分类器在人脸识别领域取得了令人惊喜的成就,它将训练样本看成字典,寻求测试样本在字典下的最稀疏的表示,即用尽可能少的训练样本的线性组合来重构测试样本。但是经典的基于稀疏表示的分类器没有考虑训练样本的类别信息,以致被选中的训练样本来自许多类,不利于分类,因此基于组稀疏的分类器被提出。组稀疏方法考虑了训练样本的类别相似性,其目的是用尽可能少类别的训练样本来表示测试样本,然而这类方法的缺点是同类的训练样本或者同时被选中或者同时被丢弃。在实际中,人脸受到光照、表情、姿势甚至遮挡等因素的影响,样本之间关系比较复杂,因此最后介绍局部加权组结构稀疏表示方法。该方法尽量用来自于与测试样本相似的类的训练样本和来自测试样本邻域的训练样本来表示测试样本,以减轻不相关类的干扰,并使得表示更稀疏和更具判别性。
简介:用分子对接和三维全息原子场作用矢量方法对36个来曲唑类衍生物和34个阿那曲唑类衍生物与芳香化酶的作用模式进行了研究,建立了三维定量构效关系模型,并在分子水平上阐述了其结合机制.运用多元线性回归(MLR)建模,同时采用内部及外部双重验证的办法对所得模型稳定性能进行深入分析和检验.MLR建模的复相关系数(Rcum)、留一法交互校验复相关系数(QCV)和外部样本校验复相关系数(Qext)分别为0.863,0.782,0.796和0.931,0.825和0.641.预测模型具有良好的稳定性和预测能力.采用AutoDock4.2软件对药物与受体之间的结合方式进行了研究.运用这些信息能为进一步设计合成强效芳香化酶抑制剂,或筛选潜在的具有更强抑制活性的天然化合物提供帮助.
简介:Singlecrystal(100)InPsamplesand(0001)GaNepitaxiallayerswereirradiatedattheHeavyIonResearchFacilityinLanzhou(HIRFL)with86Krionsatroomtemperature.Theionfluencewasvariedfrom5×1010to1×1012cm?2.Additionally,thinaluminumfoilswithdifferentthickness(sometensofmicrometers)wereplacedinfrontofsomesamplestodeceleratetheSHI's.Ionbeamscanningwasusedtoirradiatethewholesamplesurfaceinauniformwayandmaintainednormalincidence.Topreventsampleheatingduringhigh-energyirradiation,thefluxwaskeptconstantbelow1.3×1010cm?2s?1.ThemodificationsofthesampleswereinvestigatedbyXRD.
简介:Duetolowactivationcharacteristics,desirablehigh-temperaturestrength,goodresistancetoradiationdamageandusablefabricationproperties,vanadium(V)alloysareattractivecandidatestructuralmaterialsforfusionreactors[1].Irradiationinducedhardening/embrittlementatlowtemperatureisamajorproblemforthematerialsapplicationinfusionreactor[2].Inthisstudy,H/Heionswithvariousenergieswereusedtoirradiateapurevanadium(V)andaValloy(V-4Ti)toobtainadamageplateaufromsamplesurfacetothedepthof1.5m,asshowninFig.1[3].Thedetailsofirradiationparameters(energies,fluences)forHandHeionsareshowninTable1.NanoindentationwasperformedtoinvestigatethehardeningbehaviorofV-4TialloyandpureVunderirradiation.
简介:AnovelmethodformodelingcellularmaterialsisproposedbasedonMATLABimageprocessingandsynchrotronX-raycomputedtomographyscanningtoobtainanaccuratecalculationresultofaluminumfoambasedonfiniteelementmodel.Themaximumentropyalgorithmisemployedtoobtainthebinarizationimage,andthemedianfilteringalgorithmisusedtoreducethenoiseafterbinarization.Theexternalcontourandinternalporesboundaryisextractedbythe"edge"functioninMATLAB,andthegeometricalmodelisreconstructed.Atwo-stepmeshalgorithmisadoptedtomeshthereconstructedgeometricalmodel.Accordingly,thefiniteelementmodelofaluminumfoamisestablishedbytheproposedmethodbasedonreconstructiongeometricalmodel.Thecompressionbehaviorofaluminumfoamisobtainedat25℃,100℃,200℃byABAQUS,andgoodagreementswithexperimentsareachievedbyapplyingthepresentreconstructionalgorithmandmodelingmethod.
简介:Thenanomaterialsoftenexhibitveryinterestingelectrical,optical,magnetic,andchemicalproperties,whichcannotbeachievedbytheirbulkcounterparts[1-7].Thedevelopmentofuniformnanometersizedparticleshasbeenintensivelypursuedbecauseoftheirtechnologicalandfundamentalscientificimportance[8-15].Itissignificantthatnanostructuredmaterialscanbecontrollablyassembledintotherequiredgeometryontosubstrates,becomingthebasisofthenextgenerationofcomponentsanddevices[16-31].Thedevelopmentofnewmethodsandstrategiesfororganizingthenanoparticlebasicbuildingblocksintothedesiredstructuresisrequired.Superlatticesmadefromthesebuildingblocksgiveustheopportunitytostudynotonlythepropertiesoftheindividualbuildingblocks,butalsocollectiveeffects.Thesuperparamagneticironoxidenanocrystals(NCs)havebeenusedinthefieldsofbio-medicine,ferrofluids,refrigerationsystem,catalysis,particularlyinmagneticresonanceimaging,tissueengineering,anddrugdeliveryapplications[32-42].
简介:Nanowireshaveattractedmoreandmoreinterestinrecentyearsduetotheiruniquepropertiescomparedtobulkmaterials[1].Thequitedifferentpropertiespossessedbynanowiresaremainlycausedbytheirhighsurface-tovolumeratio[2].Behaviorsofnanowiresunderirradiationconditionshavealsoattractedintensiveattentionintheseyears[3??5].Ithasbeenproposedthatnanowiresmayberadiationresistancebecausetheirlargesurface-to-volumeratio,andmeanwhilethesurfacearetheperfectsinksfordefectscreatedintheradiationprocess[3].
简介:Sweetsorghum,theworld'sfifthmostimportantcerealafterrice,corn,wheatandbarley,isaveryattractivesourceforbiomassproduction.Thisrawmaterialcomplementssugarcaneproductionandisemergingastheidealfeedforanimalsandfortheproductionofthefirst-generationofbiofuels,suchasethanolandvalue-addedchemicals.Atpresent,mostofthestudiesaboutsweetsorghumfocusonutilization(suchasethanolproduction),andthestudyofsugeraccumulationinstemarerarely.
简介:Duringthepastyear,biophysicsgroupatInstituteofModernPhysics(IMP)obtainedlotsofachievementsintheresearchofheavy-ionmutationbreedingandproductionchainofsweetsorghum.Infundamentalresearchfield,amutantpopulationofArabidopsisthalianainducedbycarbonionbeamradiationwasestablishedinM2generation.Thetotalmutationratewas4.77%.Amutantmarked197#,whichhadfrostbite-like,palegreen,wrinkledandunevenleavesanddisplayedloosebractsandlatematuration,wasobtainedandreportedforthefirsttime.Generoughmappingresultsdemonstratedthatthereweretwomutationsitesinthe1stand4thchromosomeof197#mutant,indicatingheavyionradiationmightinducemorecomplicatedmutationsbeyondourcurrentrecognition.Thewholegenomeresequencingofthismutantisstillinprogress.Inaddition,high-yieldstrainsofmicrobeswhichhavepotentialvalueforcommercialapplication,suchasCorynebacteriumglutamicumandLacbobacillusthermophiles,werescreenedusingheavy-ionmutationtechnique.
简介:采用硝酸-盐酸-氢氟酸高压封闭消解样品,电感耦合等离子体原子发射光谱法测定玄武岩中的8种微量元素。实验确定了方法的分解条件以及测定元素的检出限及干扰条件。用国家一级标准物质GBW07105(玄武岩)进行方法验证,检测结果与标准值基本吻合,符合地质矿产开发的要求。
简介:Thesymmetryenergy,whichgovernsmanyphysicalphenomenafromthestructureofexoticnucleitoastrophysicalprocesses,hasmanyramificationsinbothnuclearphysicsandastrophysics.Althoughnuclearsymmetryenergyanditsslopeatthenormaldensityofnuclearmatterhavebeenroughlypinneddown,recentinterpretationsofFOPIandFOPI-LANDdatabydifferenttransportmodelsleadtodivergentconclusionsonthedensity-dependentsymmetryenergyatsupradensities.
简介:系统综述了自19世纪开始至今常用的统计相关性的方法,例如Pearson和Spearman相关系数,CorGc和CovGc相关性及距离相关性方法。重点介绍了2011年提出的MIC方法以及由此引发的毁誉参半的大量评述,旨在揭示这一热点领域的研究面貌。该领域不仅受到统计学家的关注,而且受到了分析大样本和异质数据的应用研究领域的学者们的追捧,例如基因组生物学家和网络信息研究者。这些研究者期望在众多已有方法的理解和剖析中更恰当地付诸应用,并提出新的应用问题来推动新的分析方法的创造。