中国科大学位与研究生教育
课程名称: 教师:
当前位置:
 >> 
课程列表  
主讲教师:刘天宇 人气:285 更新时间: 2017年01月09日
报告内容:蒙特卡洛模拟是一种基于抽样的数值计算方法,在医学物理中常用于辐射剂量计算。其特点是非常精准,同时也极为耗时。近年来,计算机硬件技术高速发展,以GPU为代表的硬件加速器引领了异构并行计算的新方向。美国伦斯勒理工学院的研究小组一直在探索利用GPU加速蒙卡模拟的方法,开发了快速剂量计算引擎Archer。本讲座着重介绍Archer的开发过程和初步结果,并分析目前GPU蒙特卡洛计算的主要瓶颈。主讲人简介:刘天宇博士,2009年毕业于清华大学,博士期间就读于美国伦斯勒理工学院RPI徐榭教授(千人计划)实验室,2014年毕业留校任教至今。美国核学会ANS、美国保健物理协会HPS、美国医学物理协会AAPM荣誉会员,并担任《RadiationMeasurement》《RadiationProtectionDosimetry》《MedicalPhysicsJournal》《IEEETransactionsonMedicalImaging》杂志审稿人。研究领域包括医学物理、保健物理和辐射剂量学,研究方向为快速蒙特卡洛剂量计算引擎开发,并发表多篇专著,在蒙特卡洛软件应用及开发方面有丰富经验。
主讲教师:郭卫红 人气:120 更新时间: 2017年01月09日
内容提要:Imagesuper-resolution(SR),aprocesstoenhanceimageresolution,hasimportantapplicationsinsatelliteimaging,high-definitiontelevision,medicalimaging,andsoon.Manyexistingapproachesusemultiplelow-resolution(LR)imagestorecoveronehigh-resolution(HR)image.WewillpresentaniterativeschemetosolvesingleimageSRproblems.Itrecoversahigh-qualityHRimagefromsolelyoneLRimagewithoutusingatrainingdataset.Wesolvetheproblemfromimageintensityfunctionestimationperspectiveandassumethattheimagecontainssmoothandedgecomponents.Wemodelthesmoothcomponentsofanimageusingathin-platereproducingkernelHilbertspaceandtheedgesusingapproximatedHeavisidefunctions.Wewillalsotalkabouttheextensionoftheideatopansharpeningwhichfuses3Dlowresolutionmultispectralimagesanda2Dhighresolutionpanchromaticimagesfor3Dhighresolutionmultispectralimages.TheresultsincolvecollaborationwithRichardLartey(CWRU,USA),Liang-JianDeng(UESTC,China),JocelynChanussot(GrenobleInstituteofTechnology,France)andXiaoxiangZhu(TUM,Germany).报告人简介:WeihongGuoisanAppliedMathAssociateProfessoratCaseWesternReserveUniversity,USA.ShereceivedthePh.D.degreeinAppliedMathfromUniversityofFlorida(USA)in2007.ShealsoreceivedaMaster’sdegreeinStatisticsfromthesameuniversitythesameyear.Herresearchinterestsincludeimagereconstructionandimageprocessingsuchasimagesuper-resolution,imagesegmentation,imageregistrationandtheirapplicationsinmedicine,biology,remotesensing,satelliteimagingetc.Prof.GuoisanassociateeditoroftheinternationaljournalInverseProblemsandImaging.ShealsoservedasaneditorforInternationalJournalofBiomedicalImaging.Shehaspublishedmorethan30papersinvariouswell-knowninternationaljournalsincludingSIAMJ.ImagingSciences,InverseProblemsandImaging,InformationSciences,JournalofComputationalandAppliedMathematics,IEEETransactionsonCircuitsandSystemsforVideoTechnology,MagneticResonanceImagingandMagneticResonanceinMedicine.Prof.Guohasalsoreferredpapersformorethan20internationaljournalsandconferences.
主讲教师:许封元 人气:212 更新时间: 2017年01月09日
南京大学许封元教授学术报告会题目:EfficientLarge-scaleTemporalGraphAnalyticsforCyberSecurity时间:2017年1月4日10:30地点:西区电三楼632学术报告厅
主讲教师:Xiaoyong Shen 人气:194 更新时间: 2017年01月09日
摘要:Recently,convolutionalneuralnetworks(CNNs)achievedsuccessinimageprocessingareasasaregressiontool.Inthistalk,IwillfirstreviewtherecentlydevelopedimageprocessingtechniquesusingCNNs.Then,anewCNNsframeworknamedconvolutionalneuralpyramid(CNNs)basedonournewobservationswillbeintroduced.报告人简介:XiaoyongShenisapostdoctoralfellowintheChineseUniversityofHongKong.HegotthePh.D.degreeinComputerScienceandEngineeringDepartmentintheChineseUniversityofHongKongin2016.HissupervisorisProf.JiayaJia.Beforethat,HereceivedtheB.S.degreeinComputationalMathematicsandM.S.degreeinAppliedMathematicsfromZhejiangUniversityin2010and2012respectively,underthesupervisionofProf.LigangLiu.Hisresearchinterestincludescomputergraphicsandcomputervision.
主讲教师:刘霁 人气:213 更新时间: 2016年12月30日
摘要:Thistalkconsiderstwocomputationalchallengesinmachinelearning(ML):1)HowtoimprovetheefficiencyofparallelizationforsolvinglargescaleMLproblems;and2)howtosystematicallysolvemoresophisticatedMLproblemsintheformofcompositionoptimizationsuchasreinforcementlearning,nonlineardimensionreduction,andgraphicalmodels.For1),thistalkwillintroducetheasynchronousparalleloptimization,thatopensanewgatewaytoaddressbigdatainvolvedoptimizationandanalytics.Incomparisontothetraditionalsynchronousparallelism,theasynchronousfashionsignificantlyreducesthesystemoverheadandmaximizestheefficiencyonthesystemlevelandhasbeensuccessfullyappliedtodeeplearning,recommendationsystem,NLP,highperformancecomputing,andmanyotherareasrecently.Thistalkwillintroduceseveralrecentworkdonebythespeakerabouttheasynchronousparallelalgorithmsfromtheoryfoundationstoapplication,includingconvergence,speedupproperties,andapplicationsinsolvingdeeplearning,largelinearsystems,SVM,LASSO,linearprograming,etc.For2),thistalksummarizesalargegroupofMLproblemsintothecompositionoptimization,andprovidesasystematicalgorithmforsolvingcompositionoptimization.ThetheoreticalresultsprovidefundamentalunderstandingonthecomplicationandcomplexityofthesetypeMLproblems.简历:JiLiuiscurrentlyanassistantprofessorinComputerScience,ElectricalComputerEngineering,andGoergenInstituteforDataScienceatUniversityofRochester(UR).HereceivedhisPh.D.,Masters,andB.S.degreesfromUniversityofWisconsin-Madison,ArizonaStateUniversity,andUniversityofScienceandTechnologyofChinarespectively.Hisresearchinterestscoverabroadscopeofmachinelearning,optimization,andtheirapplicationsinotherareassuchashealthcare,bioinformatics,computervision,andmanyotherdataanalysisinvolvedareas.Hisrecentresearchfocusisonasynchronousparalleloptimization,sparselearning(compressedsensing)theoryandalgorithm,reinforcementlearning,structuralmodelestimation,onlinelearning,abnormaleventdetection,feature/patternextraction,etc.HefoundedthemachinelearningandoptimizationgroupatUR.HewontheawardofBestPaperhonorablementionatSIGKDD2010andtheawardofFacebookBestStudentPaperawardatUAI2015.
主讲教师:苗付友 人气:1089 更新时间: 2016年12月21日
2016年秋季课程
主讲教师:信息 人气:2161 更新时间: 2016年12月21日
2016年秋季课程
主讲教师:彭春翌 人气:295 更新时间: 2016年12月19日
报告人:彭春翌助理教授(美国俄亥俄州立大学)题目:HarnessEndIntelligenceofMobileDevices时间:2016年12月17日(周六)14:30地点:西区电三楼632学术报告厅
主讲教师:杨亚玲 人气:247 更新时间: 2016年12月16日
报告人:杨亚玲副教授(美国弗吉尼亚理工大学)题目:SecurityandPrivacyIssuesinDynamicSpectrumAccessNetworks时间:2016年12月15日(周四)14:00地点:西区电三楼632学术报告厅
主讲教师:熊辉 人气:595 更新时间: 2016年12月16日
美国罗格斯-新泽西州立大学熊辉教授学术报告会题目:TheRoadtoaPh.D.时间:2016年12月15日上午10:00地点:西区科技实验西楼117报告厅Abstract:Inthistalk,Iwouldliketointroduceinsightsintostrategiesforhowtohavemoreenergy,bemoreefficient,andbemorerelaxed,andgetalotmoreaccomplishedontheroadtoaPhD.Specifically,IwillsharemyexperienceaboutsomeofthemisperceptionthatmanyChinesestudentshaveduringtheirPhDstudyinUSA.Inparticular,Iwilldiscussissues,suchas“Howtobeaself-motivatedstudent?”,“WhatistherightattitudeforPhDstudy?”,“Howtobebalancedbetweenworkandlife?”,“Howimportantarecommunicationskills?”,“Howtodoconflictmanagement?”,“Howtodotimemanagement?”,“Howimportantaremarketingskills?”,“Whatisthedifferencebetweenleadershipandmanagement?”,and“Howtobuildandmaintainrelationships?”.Dr.HuiXiongiscurrentlyaFullProfessorandtheViceChairintheManagementScienceandInformationSystemsDepartment,andtheDirectorofRutgersCenterforInformationAssurance,attheRutgers,theStateUniversityofNewJersey,wherehereceivedatwo-yearearlypromotion/tenure(2009),theRutgersUniversityBoardofTrusteesResearchFellowshipforScholarlyExcellence(2009),theICDM-2011BestResearchPaperAward(2011),aRBSDean’sResearchProfessorAward(2016),andanIBMESAInnovationAward(2008).Dr.XiongisalsoaDistinguishedVisitingProfessor(GrandMasterChairProfessor)attheUniversityofScienceandTechnologyofChina(USTC).Forhisoutstandingcontributionstodataminingandmobilecomputing,hewaselectedanACMDistinguishedScientistin2014.Dr.XiongreceivedhisPh.D.inComputerSciencefromtheUniversityofMinnesota(UMN),USA,in2005,theB.E.degreeinAutomationfromtheUniversityofScienceandTechnologyofChina(USTC),Hefei,China,andtheM.S.degreeinComputerSciencefromtheNationalUniversityofSingapore(NUS),Singapore.Hisgeneralareaofresearchisdataandknowledgeengineering,withafocusondevelopingeffectiveandefficientdataanalysistechniquesforemergingdataintensiveapplications.Hehaspublishedprolificallyinrefereedjournalsandconferenceproceedings(4books,70+journalpapers,and130+conferencepapers.HeisanAssociateEditorofIEEETransactionsonDataandKnowledgeEngineering(TKDE),IEEETransactiononBigData(TBD),ACMTransactionsonKnowledgeDiscoveryfromData(TKDD),andACMTransactionsonManagementInformationSystems(TMIS).Hehasservedregularlyontheorganizationandprogramcommitteesofnumerousconferences,includingasaProgramCo-ChairoftheIndustrialandGovernmentTrackforthe18thACMSIGKDDInternationalConferenceonKnowledgeDiscoveryandDataMining(KDD),aProgramCo-Chairforthe2013IEEEInternationalConferenceonDataMining(ICDM-2013),andaGeneralCo-Chairforthe2015IEEEInternationalConferenceonDataMining(ICDM-2015).
共414条记录,每页显示10条,到第页 [首页] [上一页] [下一页] [尾页]
课程搜索
类  型:
学科方向:
学科分类:
院  系:
内  容:
教  师:
讲座专家
Aranda Pino
Profesor of Department of Algebra, Geometry and Topology Science Faculty, University of Málaga, Spain. Research Interests: Leavitt path
Nikos Hadjichristidis
Prof. Nikos Hadjichristidis, Department of Chemistry, University of Athens.
蔡荣根
中国科学院理论物理所研究员,博士生导师。中国引力和相对论天体物理学会副理事长(2008-) 中国科学技术大学交叉学科理论研究中心兼职教授
欧阳钟灿
欧阳钟灿,中国科学院院士。1968年清华大学自控系毕业,1981年清华大学物理系固体物理专业获硕士学位,1984年获光学专业理学博士学位。1985-1986年在理论物理所
Manuela M. Veloso
Manuela M. Veloso教授是国际人工智能和自主机器人领域的主要学界领导人和学术带头人之一,取得了一系列杰出成就,其中部分成果产生了深远的影响。
庄小威
庄小威,华裔美籍生物物理学家,美国国家科学院院士, 哈佛大学化学与化学生物、物理学教授,创办有庄小威实验室。
课件总访问人次:7450798
中国科学技术大学研究生网络课堂试运行版,版权属于中国科学技术大学研究生院。
本网站所有内容属于中国科学技术大学,未经允许不得下载传播。
地址:安徽省合肥市金寨路96号;邮编:230026。TEL:+86-551-63602922;E-mail:wlkt@ustc.edu.cn。