简介：说出的比赛是为一种语言或一个通讯系统的自我组织的出现的nonequilibrium动力学的一个模型。我们学习扬声器与与exp成正比的可能性从它的库存在选择一个词的最小的说出游戏的一个修改版本(R*一)，在R是名字的成功比率的地方并且一是一个悦耳的参数。由调查效果一上为方形的格子和没有规模的网络的进化过程，我们发现集中时间与增加减少一上二联网，它显示成功的词的优先的选择能加速到达一致。更有趣地，为一>0，我们发现在集中之间的关系预定并且一展览一种幂定律形式。[从作者抽象]

简介：Speakervariabilityisanimportantsourceofspeechvariationswhichmakescontinuousspeechrecognitionadifficulttask.Adaptingautomaticspeechrecognition(ASR)modelstothespeakervariationsisawell-knownstrategytocopewiththechallenge.AlmostallsuchtechniquesfocusondevelopingadaptationsolutionswithintheacousticmodelsoftheASRsystems.Althoughvariationsoftheacousticfeaturesconstituteanimportantportionoftheinter-speakervariations,theydonotcovervariationsatthephoneticlevel.Phoneticvariationsareknowntoformanimportantpartofvariationswhichareinfluencedbybothmicro-segmentalandsuprasegmentalfactors.Inter-speakerphoneticvariationsareinfluencedbythestructureandanatomyofaspeaker'sarticulatorysystemandalsohis/herspeakingstylewhichisdrivenbymanyspeakerbackgroundcharacteristicssuchasaccent,gender,age,socioeconomicandeducationalclass.Theeffectofinter-speakervariationsinthefeaturespacemaycauseexplicitphonerecognitionerrors.Theseerrorscanbecompensatedlaterbyhavingappropriatepronunciationvariantsforthelexiconentrieswhichconsiderlikelyphonemisclassificationsbesidespronunciation.Inthispaper,weintroducespeakeradaptivedynamicpronunciationmodels,whichgeneratedifferentlexiconsforvariousspeakerclustersanddifferentrangesofspeechrate.Themodelsarehybridsofspeakeradaptedcontextualrulesanddynamicgeneralizeddecisiontrees,whichtakeintoaccountwordphonologicalstructures,rateofspeech,unigramprobabilitiesandstresstogeneratepronunciationvariantsofwords.EmployingthesetofspeakeradapteddynamiclexiconsinaFarsi(Persian)continuousspeechrecognitiontaskresultsinworderrorratereductionsofasmuchas10.1%inaspeaker-dependentscenarioand7.4%inaspeaker-independentscenario.

简介：Inthispaper,wedescribetheestimationoflow-altituderefractivitystructurefromsimulationandrealground-basedGPSdelays.Theverticalstructureoftherefractiveenvironmentismodeledusingthreeparameters,i.e.,ductheight,ductthickness,andductslope.Therefractivitymodelisimplementedwithaprioriconstraintsontheductheight,thickness,andstrength,whichmightbederivedfromsoundingsornumericalweather-predictionmodels.Araypropagationmodelmapstherefractivitystructureintoareplicafield.Replicafieldsarecomparedwiththesimulationobserveddatausingasquarederrorobjectivefunction.Aglobalsearchforthethreeenvironmentalparametersisperformedusingageneticalgorithm.Theinversionisassessedbycomparingtherefractivityprofilesfromtheradiosondestothoseestimated.Thistechniquecouldprovidenear-real-timeestimationoftheductingeffect.Theresultssuggestthatground-basedGPSprovidessignificantatmosphericrefractivityinformation,despitecertainfundamentallimitationsofground-basedmeasurements.Radiosondesaretypicallylaunchedjustafewtimesdaily.Consequently,estimatesoftemporallyandspatiallyvaryingrefractivitythatassimilateGPSdelayscouldsubstantiallyimproveover-estimatescausedbyusingradiosondedataalone.

简介：针对飞行器高动态飞行环境，对GPS应用的几点关键问题进行了分析和研究，并且利用仿真建模平台对研究结果进行仿真验证。首先根据再入航天器飞行高度高、速度快、飞行距离远等特点，对GPS定位方式进行选择。采用了伪距单点绝对定位方法进行定位，硬件要求简单、效费比高，能达到预期的精度，由此对伪距单点定位测速方法进行了介绍。接着结合载体高速运动的特性，在分析了通常的电离层余弦修正模式和介绍电离层的基本特性与描述的基础上，提出了动态修正的定位方法。该方法是以Bent电离层模型为基础，

简介：

简介：

简介：

简介：

简介：