资源描述:
《Machine Learning and Understanding机器学习与理解》由会员上传分享,免费在线阅读,更多相关内容在教育资源-天天文库。
1、MachineLearningandUnderstandingforIntelligentExtremeScaleScientificComputingandDiscovery15-CS-1768DOEWorkshopReportJanuary5-7,2015Rockville,MDASCRMachineLearningforExtremeScaleComputingWorkshopReportiMachineLearningandUnderstandingforIntelligentExtremeScaleScientificComputin
2、gandDiscoveryDOEWorkshopReportJanuary7–9,2015Rockville,MDWorkshopOrganizingCommitteeMichaelBerry(Chair),UniversityofTennesseeThomasE.Potok,OakRidgeNationalLaboratoryPrasannaBalaprakash,ArgonneNationalLaboratoryHankHoffmann,UniversityofChicagoRajuVatsavai,NorthCarolinaStateUn
3、iversityPrabhat,LawrenceBerkeleyNationalLaboratoryDOEASCRPointofContactRobinsonPinoCover:MachinelearningtechniquescanbeappliedtoawiderangeofDOEresearchareas,suchasautomaticallyidentifyingweatherphenomenainmassivesimulationdatasets..ASCRMachineLearningforExtremeScaleComputing
4、WorkshopReportContents1ExecutiveSummary11.1Self-AwareOperatingandRuntimeSystems11.2MachineLearning21.3ResilienceandTrust22Acknowledgements42.1Self-AwareOperatingandRuntimeSystems42.2DeepLearning42.3ResilienceandTrust43Introduction43.1Self-AwareOperatingandRuntimeSystems43.2I
5、ntroductiontoMachineLearning53.3ResilienceandTrust84MotivatingScience104.1DOEScienceDrivers104.2Self-AwareOperatingandRuntimeSystems184.3ExampleofFutureOS/R195MachineLearning215.1MachineLearning215.2DataCollection,ManagementandIntegration225.3Metrics235.4MotivatingSciencefro
6、mTopic3246ChallengesofMachineUnderstandingandLearning256.1ChallengesofMLforScientificDiscoveryfromTopic2256.2ChallengesofMLforHighPerformanceComputing286.3ChallengesatExascalefromTopic3316.4ChallengesforMLtowardsResilienceandTrustfromTopic3317CurrentandFutureResearchDirectio
7、ns337.1ResearchDirectionsinMLfromTopic2337.2CurrentandFutureResearchDirectionsfromTopic3358InterdependencieswithOtherEfforts478.1InterdependencieswithOtherEffortsfromTopic2479InterdependencieswithOtherEffortsfromTopic34710CommonThemes,Findings,andRecommendations4810.1CommonT
8、hemes,FindingsandRecommendationsfromTopic24810.2CommonThemes,Findings,andRe