《人工智能-数据科学比赛汇总 20197.docx》由会员分享，可在线阅读，更多相关《人工智能-数据科学比赛汇总 20197.docx（40页珍藏版）》请在taowenge.com淘文阁网|工程机械CAD图纸|机械工程制图|CAD装配图下载|SolidWorks_CaTia_CAD_UG_PROE_设计图分享下载上搜索。
1、人工智能/数据科学比赛汇总2019.7Note:ARIEL,amissiontomakethefirstlarge-scalesurveyofexoplanetatmospheres,haslaunchedaglobalcompetitionseriestofindinnovativesolutionsfortheinterpretationandanalysisofexoplanetdata.Youcanfindourpressrelease.invitesprofessionalandamateurdatascientistsaroundtheworldtouseMachineLearni
2、ng(ML)toremovenoisefromexoplanetobservationscausedbystarspotsandbyinstrumentation.thatfocusesontheretrievalofspectrafromsimulationsofcloudyandcloud-freesuper-Earthandhot-Jupiterdatawasalsolaunchedtoday.tocreatepipelinesforfaster,moreeffectiveprocessingoftherawdatagatheredbythemissionwillbelaunchedin
3、June.EntryDeadline:2nd3DFaceAlignmentintheWildChallenge-DenseReconstructionfromVideoJuly4-Aug152019/Hostby/Prize:NaNNote:The2nd3DFAWChallengeevaluates3Dfacereconstructionmethodsonanewlargecorporaofprofile-to-profilefacevideosannotatedwithcorrespondinghigh-resolution3Dgroundtruthmeshes.Thecorporaincl
4、udesprofile-to-profilevideosobtainedunderarangeofconditions:high-definitionin-the-labvideo,unconstrainedvideofromaniPhonedeviceEntryDeadline:飯田産業土地販売価格推定6月10日-8月2019年度/Hostby/Prize:2,300,000Note:日本語話方EntryDeadline:AI开发者大赛5月21日-9月21日,2019/Hostby/Prize:1000000x8Note:EntryDeadline:AutoCV2:Imageandvideo
5、ClassificationJuly2-Aug20,2019/Hostby/Prize:4000USDNote:Thisisround2ofAutoCV:ImageVideo!Thisisa2-phasechallenge,seethechallengerulesfordetails.ThisistheFEED-BACKPHASE.Thesecondphase(finalblind-testphase)willberunfromaseparatesubmissionsite,tobeannouncedaftertheendofthefeed-backphase.EntryDeadline:iF
6、LYTEKAI开发者大赛5月21日-10月14日,2019/Hostby/Prize:100万RMBNote:iFLYTEKAI开发者大赛是由科大讯飞提议的顶尖人工智能竞赛平台会聚产学研各界力量面向全球开发者提议数据算法及创新应用类挑战推动人工智能前沿科学研究以及创新成果转化培育人工智能产业人才助力人工智能生态建立。2019年度第二届iFLYTEKAI开发者大赛将继续开放科大讯飞优质大数据资源及人工智能核心技术面向全球开发者提议数据算法及创新应用类挑战。阿尔茨海默综合症预测挑战赛:基于老年度人在特定图片描绘任务中产生的语音给定语音数据中提取出的声学特征、主被试对话的切分信息、人工文本转写结果和
7、对应的认知标签建立2分类模型预测认知标签正常或者认知障碍。挪动广告反欺诈算法挑战赛:挪动广告反欺诈需要强大的数据作为支撑本次大赛提供了讯飞AI营销云海量的现网流量数据作为训练样本参赛选手需基于提供的样本构建模型预测流量作弊与否。大数据应用分类标注挑战赛:选手基于提供的应用二级分类标签和假设干随机应用标注样本实现应用分类标注算法每个应用一个标签以应用最主要属性对应的标签为该应用的标签。工程机械核心部件寿命预测挑战赛:由中科云谷科技有限公司提供某类工程机械设备的核心耗损性部件的工作数据包括部件工作时长、转速、温度、电压、电流等多类工况数据。祈望参赛者利用大数据分析、机器学习、深度学习等方法提取适宜
8、的特征、建立适宜的寿命预测模型预测核心耗损性部件的剩余寿命。EntryDeadline:SIIM-ACRPneumothoraxSegmentationNow-Sept22,2019/Hostby/Prize:$30,000Note:IdentifyPneumothoraxdiseaseinchestx-raysEntryDeadline:PredictingMolecularPropertiesNow-August28,2019/Hostby/Prize:$30,000Note:Canyoumeasurethemagneticinteractionsbetweenapairofatoms?E
9、ntryDeadline:APTOS2019BlindnessDetectionNow-Sept5,2019/Hostby/Prize:$50,000Note:DetectdiabeticretinopathytostopblindnessbeforeitstoolateEntryDeadline:QMULSurveillanceFaceRecognitionChallengeICCV2019workshopRLQ27June-30Aug,2019/Hostby/Prize:NaNNote:Thechallengedataconsistsofasetofpopularsearchqueries
10、andafairsizesetofcandidatedocuments.Challengeparticipantsmakeabooleanrelevant-or-notdecisionforeachquery-documentpair.Humanjudgmentsareusedtocreatelabeledtrainingandevaluationdataforasubsetofthequery-documentpairs.EvaluationofsubmissionswillbebasedonthetraditionalF1metric,incorporatingcomponentsofbo
11、threcallandprecision.EntryDeadline:“达观杯文本智能信息抽取挑战赛06/28-08/312019/Hostby/Prize:七万七千元Note:本次大赛的任务是给定一定数量的标注语料和海量的未标注语料在3个字段上做信息抽取任务。EntryDeadline:ChallengeonDeepLearningbasedLoopFilterforVideoCodingMay,25th-May,31st,2018/Hostby/Prize:NaNNote:Theparticipantsareencouragedtoinvestigateneuralnetworkbased
12、methods(especiallyconvolutionalneuralnetworks)withdifferentnetworkstructures,inahopeofachievingthebestqualitywithlightestnetworkconfigurationforagoodtradeoffofefficiencyandcomplexity.EntryDeadline:CoNLL2019ShardTaskonCross-FrameworkMeaningRepresentationParsingMarch6-November3,2019/Hostby/Prize:NaNNo
13、te:The2019ConferenceonComputationalLanguageLearning(CoNLL)hostsasharedtask(orsystembake-off)onCross-FrameworkMeaningRepresentationParsing(MRP2019).Thegoalofthetaskistoadvancedata-drivenparsingintograph-structuredrepresentationsofsentencemeaning.EntryDeadline:The2ndLarge-scaleVideoObjectSegmentationC
14、hallengeMay.20-Sep.52019/Hostby/Prize:NaNNote:Asacontinuousefforttopushforwardtheresearchonvideoobjectsegmentationtasks,weplantohostasecondworkshopwithachallengebasedontheYouTube-VOSdataset,targetingatmorediversifiedproblemsettings,i.e.,weplantoprovidetwochallengetracksinthisworkshop.EntryDeadline:T
15、he3rdYouTube-8MVideoUnderstandingChallengeNow-October28,2019/Hostby/Prize:$25,000Note:TemporallocalizationoftopicswithinvideoEntryDeadline:AutomaticStructureSegmentationforRadiotherapyPlanningChallenge2019June15-Oct1,2019/Hostby/Prize:NaNNote:Thegoalofthechallengeistosetuptasksforevaluatingautomatic
16、algorithmsonsegmentationoforgans-at-risk(OAR)andgrosstargetvolume(GTV)oftumorsoftwotypesofcancers,nasopharynxcancerandlungcancer,forradiationtherapyplanning.Therearefourtasksforevaluatingtheperformanceofthealgorithms.Participantscanchoosetojoinalloreithertasksaccordingtotheirinterests.Task1:Organ-at
17、-risksegmentationfromheadneckCTscans.Task2:GrossTargetVolumesegmentationofnasopharynxcancer.Task3:Organ-at-risksegmentationfromchestCTscans.Task4:GrossTargetVolumesegmentationoflungcancer.EntryDeadline:OpenEDSChallengeMay3-Sep16,2019/Hostby/Prize:$13,000USDx2Note:Intheabsenceofaccurategazelabels,wep
18、roposetoadvancethestateoftheartbycarefullydesigningtwochallengesthatcombinehumanannotationofeyefeatureswithunlabeleddata.Thesechallengesfocusondeeperunderstandingofthedistributionunderlyinghumaneyestate.WeinviteMLandCVresearchersforparticipation.EntryDeadline:ExoplanetimagingdatachallengeMay16th-Sep
19、16th,2019/Hostby/Prize:NaNNote:Thiscompetitioniscomposedoftwosub-challengesfocusingonthetwomostwidelyusedobservingtechniques:pupiltracking(angulardifferentialimaging,ADI)andmulti-spectralimagingcombinedwithpupiltracking(multi-channelspectraldifferentialimaging,ADImSDI).EntryDeadline:成语浏览理解大赛06/25-09
20、/252019/Hostby/Prize:24,000元Note:本次竞赛将基于选词填空的任务形式提供大规模的成语填空训练语料。在给定假设干段文本下选手需要在提供的候选项中依次选出填入文本中的空格处最恰当的成语。EntryDeadline:PekingUniversityInternationalCompetitiononOcularDiseaseIntelligentRecognition(ODIR-2019)May18-Sep25,2019/Hostby/Prize:10,00,000RMB(140,000USD)Note:北京大学智慧之眼国际眼科疾病智能识别竞赛TheSGwillprov
21、ideparticipantswith5,000structureddesensitizedophthalmologicimagesetofpatientsage,sex,binocularcolorfundusphotosanddoctorsdiagnosticreport.上工医信将为参赛者提供5000组包含患者的性别、年度龄、双眼彩色眼底照片以及医生印象报告等的构造化脱敏后眼科的数据集。Thepurposeofthischallengeistocompareapproachesofophthalmicdiseaseclassificationincolorfundusimages.Par
22、ticipantwillhavetosubmitclassificationresultsofeightcategoriesforallthetestingdata.Foreverycategory,aclassificationprobability(valuefrom0.0to1.0)denotesriskofapatientdiagnosedwithcorrespondingcategory.该竞赛的目的是比拟基于彩色眼底图像进展眼科疾病分类的不同方法。介入者必须提交所有测试数据集的八个类别的分类结果。对于每个类别分类概率值从0.0到1.0表示患者被诊断为具有相应类别的可能性/风险。En
23、tryDeadline:Digestive-SystemPathologicalDetectionandSegmentationChallenge2019June14-Oct1,2019/Hostby/Prize:NaNNote:Thegoalofthechallengeistosetuptasksforevaluatingautomaticalgorithmsonsignetringcelldetectionandcolonoscopytissuescreeningfromdigestivesystempathologicalimages.Thiswillbethefirstchalleng
24、eandfirstpublicdatasetonsignetringcelldetectionandcolonoscopytissuescreening.Releasingthelargequantityofexpert-levelannotationsondigestive-systempathologicalimageswillsubstantiallyadvancetheresearchonautomaticpathologicalobjectdetectionandlesionsegmentation.Task1:Signetringcelldetection.Task2:Colono
25、scopytissuesegmentationandclassification.EntryDeadline:The2ndChina(Hengqin)InternationalUniversityQuantitativeFinanceCompetition2019-04-19至2020-03-21/Hostby/Prize:140万Note:第二届中国横琴国际高校量化金融大赛参赛要求参赛者应根据题目要求完成一篇包括量化金融策略原理、模型的假设、建立以及求解、计算方法的设计、分析以及检验、模型的改良等方面的书面报告即答卷并在规定竞赛期间内将参赛策略的市场运行进展模拟仿真竞赛。根据参赛策略的测试结
26、果包括样本内以及样本外的收益程度及市场风险防范的效果等统一指标打分评比以市场的标准来决定优劣评价策略的回测以及实盘模拟表现同时考虑策略逻辑的稳健性以及创新性。竞赛评奖以策略的合理性、建模的创新性、测试策略的市场适应性及收益风险程度等结果为主要标准。RequirementsParticipantsshouldwriteareportcoveringquantitativefinancialstrategytheories1)Modeltheoreticalhypothesisanddescriptionofquantitivemodel2)Dataanalysis3)Strategybackte
27、stingresultsandperformanceanalysis.Accordingtotherequirementsofthecompetition,participantsstrategieswillbebacktestedandpapertradedduringtherequiredperiod.Evaluationandscoringwillbaseonunifiedmeasurementsincludingreturn,volatility,maxdrawdownofthestrategiesandsoon.Thedeterminationofmeritsandevaluatio
28、nofstrategybacktestandpapertradingperformancewillbemadeaccordingtomarketstandards,whiletherobustnessandinnovationofthestrategiclogicwillalsobetakenintoconsideration.Keycriteriawillincludetherationalityofthestrategy,thecreativenessofthemodel,themarketadaptabilityofthetestingstrategyandthelevelofretur
29、nandrisk.EntryDeadline:OpenImages2019Now-Oct27,2019/Hostby/Prize:$25,000Note:Thisyearsreleaseenabledthesecondtoincludethefollowing3tracks:trackfordetectingboundingboxesaroundobjectinstances,relaunchedfrom2018.fordetectingpairsofobjectsinparticularrelations,alsorelaunchedfrom2018.Instancesegmentation
30、trackLinktobeprovidedwhenlaunchedonJuly1,brandnewfor2019.EntryDeadline:VisualDomainAdaptationChallenge(VisDA-2019)April9-Sept.27,2019/Hostby/Prize:NaNNote:Wearepleasedtoannouncethe2019VisualDomainAdaptation(VisDA2019)Challenge!Itiswellknownthatthesuccessofmachinelearningmethodsonvisualrecognitiontas
31、ksishighlydependentonaccesstolargelabeleddatasets.Unfortunately,performanceoftendropssignificantlywhenthemodelispresentedwithdatafromanewdeploymentdomainwhichitdidnotseeintraining,aproblemknownasdatasetshift.TheVisDAchallengeaimstotestdomainadaptationmethodsabilitytotransfersourceknowledgeandadaptit
32、tonoveltargetdomains.Thischallengeincludestwotracks:EntryDeadline:AIinRTC-超分辨率图像质量比拟挑战赛7月1日-10月23日,2019/Hostby/Prize:100000Note:单帧图像超分辨率近年度来备受关注。同样的图像在经过不同超分辨率算法处理后获得的图像质量也有所不同。在这个挑战中参赛者需要对100张图片进展4倍超分辨率处理。比赛最终以PI(perceptualindex)指标作为评判标准PI值越小说明图像质量越高得分越高分值高的团队获得优胜。EntryDeadline:AIinRTC-超分辨率算法性能比拟挑战
33、赛7月1日-10月23日,2019/Hostby/Prize:100000Note:将超分辨算法用于处理实时视频流时模型的处理表现与运算性能是一个两难的选择。为了追求较低复杂度可能需要牺牲图像质量为了追求较高质量的输出导致设备资源占用过高产生设备发烫、视频模糊卡顿等现象。该挑战主要考察算法模型的性能参赛者需要对图像做2倍的超分辨率处理算法复杂度控制在1GFLOPs之内我们以SRCNN模型为baseline,并采用PSNR、SSIM及运行时间来综合评估算法的性能分值高者即获胜。EntryDeadline:AlchemyContest5/22-9/30,2019/Hostby/Prize:tota
34、l100,000RMBNote:TheTencentQuantumLabhasrecentlyintroducedanewmoleculardataset,calledAlchemy,tofacilitatethedevelopmentofnewmachinelearningmodelsusefulforchemistryandmaterialsscience.Thedatasetlists12quantummechanicalpropertiesof130,000organicmoleculescomprisingupto12heavyatoms(C,N,O,S,FandCl),sample
35、dfromthedatabase.Thesepropertieshavebeencalculatedusingtheopen-sourcecomputationalchemistryprogramPython-basedSimulationofChemistryFramework().EntryDeadline:FashionIQChallengeJune1-Sept30,2019/Hostby/Prize:NaNNote:FashionIQisanewdatasetwecontributetotheresearchcommunitytofacilitateresearchonnaturall
36、anguagebasedinteractiveimageretrievalEntryDeadline:MicroNetChallengeNeurIPS2019June1,2018-Dec13,2019/Hostby/Prize:NaNNote:Thecompetitionconsistsofthreedifferenttasks.Contestantsarefreetosubmitentriesforone,two,orallthreetasks.Contestantsareallowedtoenteruptothreemodelsforeachtask,butwillberankedacco
37、rdingtotheirtopentryineachtask.Entriescanonlybetrainedonthetrainingdataforthetasktheyareenteredin.Nopre-training,oruseofauxiliarydataisallowed.:Thedefactostandarddatasetforimageclassification.Thedatasetiscomposedof1,281,167trainingimagesand50,000developmentimages.Entriesarerequiredtoachieve75%top-1a
38、ccuracyonthepublictestset.:Awidelypopularimageclassificationdatasetofsmallimages.Thedatasetiscomposedof50,000trainingimagesand10,000developmentimages.Entriesarerequiredtoachieve80%top-1accuracyonthetestset.:Alanguagemodelingdatasetthatemphasizeslong-termdependencies.Entrieswillperformthestandardlang
39、uagemodelingtask,predictingthenexttokenfromthecurrentone.Thedatasetiscomposedof103milliontrainingwords,217thousanddevelopmentwords,and245thousandtestingwords.Entriesshouldusethestandardword-levelvocabularyof267,735tokens.Entriesarerequiredtoachieveaword-levelperplexitybelow35onthetestset.EntryDeadli
40、ne:Multi-domainTask-CompletionDialogChallengeDSTC8June17-Oct6,2019/Hostby/Prize:NaNNote:AspartoftheEighthDialogSystemTechnologyChallenge(),MicrosoftResearchandTsinghuaUniversityarehostingatrackintendedtofosterprogressintwoimportantaspectsofdialogsystems:dialogcomplexityandscalingtonewdomains.Forthis
41、DSTC8track,therearetwotasksyoucancompetein(seebelow).ThechallengerunsfromJune17,2019October6,2019.Participantswillbuildandialogsystemfortouristinformationdesksettings.Participantswilldevelopmethodsforbuildingaconversationmodelthatgeneratesappropriatedomain-specificuserresponsestoanincompletedialoghi
42、story.EntryDeadline:EndoscopicVisionChallenge2019June5-Oct13,2019/Hostby/Prize:NaNNote:AsavisionCAIchallengeatMICCAI,ouraimistoprovideaformalframeworkforevaluatingthecurrentstateoftheart,gatherresearchersinthefieldandprovidehighqualitydatawithprotocolsforvalidatingendoscopicvisionalgorithms.EndoVis2
43、019Sub-challenges:EntryDeadline:GraphGolf:TheOrder/degreeProblemCompetition05-1311-26,2019/Hostby/Prize:NaNNote:Findagraphthathassmallestdiameteraverageshortestpathlengthgivenanorderandadegree.GraphGolfisaninternationalcompetitionoftheorder/degreeproblemsince2021.Itisconductedwiththegoalofmakingacat
44、alogofsmallest-diametergraphsforeveryorder/degreepair.Anyoneintheworldcantakepartinthecompetitionbysubmittingagraph.OutstandingauthorsareawardedinCANDAR2019,aninternationalconferenceheldinNagasaki,Japan,inNovember2019.EntryDeadline:TheAnimal-AIOlympicsApril-December2019/Hostby/Prize:$10,000Note:基于Un
45、ityMLAgentsToolkit的动物认知-AI挑战ThiscompetitionpitsourbestAIapproachesagainsttheanimalkingdomtodetermineifthegreatsuccessesofAIarenowreadytocompetewiththegreatsuccessesofevolutionattheirowngame.EntryDeadline:EPIC-KitchensActionAnticipationJuly3,2018-Nov.222019/Hostby/Prize:NaNNote:Thelargestdatasetinfir
46、st-person(egocentric)vision;multi-facetednon-scriptedrecordingsinnativeenvironments-i.e.thewearershomes,capturingalldailyactivitiesinthekitchenovermultipledays.EntryDeadline:GeopoliticalForecastingGFChallenge2April4,2018-Feb.1,2020/Hostby/Prize:$250,000Note:Solvers,whetherindividualsorteams,willcrea
47、teinnovativesolutionsandmethodstoproduceforecaststoasetofmorethan300questionsreferredtoasIndividualForecastingProblems(IFPs),releasedregularlyoverthecourseofthenine-monthChallenge.EntryDeadline:ModaNetFashionUnderstandingChallengeOct1,2018-Dec11,2019/Hostby/Prize:NaNNote:Inthischallenge,weevaluatemo
48、delperformanceforthreetasks,objectdetection,semanticsegmentationandinstancesegmentation.Youcanparticipatealltasksoranyoneofthembychoosingwhichresultstobeincludedinyoursubmission.EntryDeadline:二分类算法提供银行精准营销解决方案|练习赛2018年度12月29日-2019年度12月29日/Hostby/Prize:NaNNote:本练习赛的数据选自UCI机器学习库中的银行营销数据集(BankMarketingDataSet)EntryDeadline:SPIE-AAPM-NCIBreastPathQ:CancerCellularityChallenge2019Oct.15,2018-Dec.31,2019/Hostby/Prize:NaNNote:Participantswillbetaskedtodevelopanautomatedmethodforanalyzinghistol