A General Method for Estimating Correlated

A General Method for Estimating Correlated

ID:41634337

大小:370.16 KB

页数:12页

时间:2019-08-29

A General Method for Estimating Correlated_第1页
A General Method for Estimating Correlated_第2页
A General Method for Estimating Correlated_第3页
A General Method for Estimating Correlated_第4页
A General Method for Estimating Correlated_第5页
资源描述:

《A General Method for Estimating Correlated》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库

1、2012IEEE28thInternationalConferenceonDataEngineeringAGeneralMethodforEstimatingCorrelatedAggregatesoveraDataStreamSrikantaTirthapuraDavidP.WoodruffDept.ofElectricalandComputerEngg.IBMAlmadenResearchCenterIowaStateUniversity,Ames,IA,USASanJose,CA,USAsnt@iastate

2、.edudpwoodru@us.ibm.comAbstract—Onastreamoftwodimensionaldataitems(x,y)aggregatesarisenaturallyinanalyticsonmulti-dimensionalwherexisanitemidentifier,andyisanumericalattribute,astreamingdata,andspace-efficientmethodsforimplementingcorrelatedaggregatequeryrequire

3、sustofirstapplyaselectionsuchaggregatesareusefulinstreaminganalyticssystemssuchpredicatealongthesecond(y)dimension,followedbyanaggrega-asintheIBMSystemS[5].tionalongthefirst(x)dimension.Forselectionpredicatesoftheform(yc),whereparametercisprovidedatquery

4、Oursummariesforcorrelatedaggregatesallowmoreflexibletime,wepresentnewstreamingalgorithmsandlowerboundsforinterrogationonthedatastreamthanispossiblewithatradi-estimatingstatisticsoftheresultingsubstreamofelementsthattionalstreamsummary.Forexample,consideradatast

5、reamsatisfythepredicate.ofIPflowrecords,suchasthoseoutputbynetworkroutersWeprovidethefirstsublinearspacealgorithmsforalargeequippedwithCisco’sNetflow[6].Supposethatwewerefamilyofstatisticsinthismodel,includingfrequencymoments.Weexperimentallyvalidateouralgorithms

6、,showingthattheirinterestedinonlytwoattributesperflowrecord,thedestinationmemoryrequirementsaresignificantlysmallerthanexistingaddressoftheflow,andthesize(numberofbytes)oftheflow.linearstorageschemesforlargedatasets,whilesimultaneouslyUsingasummaryforcorrelatedagg

7、regateAGG,alongwithaachievingfastper-recordprocessingtime.quantilesummaryfortheydimension(anyofthewellknownWealsostudytheproblemwhentheitemshaveweights.streamquantilesummariescanbeusedhere,including[2],Allowingnegativeweightsallowsforanalyzingvalueswhichoccuri

8、nthesymmetricdifferenceoftwodatasets.Wegivea[7]),itispossibleforanetworkadministratortoexecutethestrongspacelowerboundwhichholdsevenifthealgorithmfollowingsequenceofqueriesonthestr

当前文档最多预览五页,下载文档查看全文

此文档下载收益归作者所有

当前文档最多预览五页,下载文档查看全文
温馨提示:
1. 部分包含数学公式或PPT动画的文件,查看预览时可能会显示错乱或异常,文件下载后无此问题,请放心下载。
2. 本文档由用户上传,版权归属用户,天天文库负责整理代发布。如果您对本文档版权有争议请及时联系客服。
3. 下载前请仔细阅读文档内容,确认文档内容符合您的需求后进行下载,若出现内容与标题不符可向本站投诉处理。
4. 下载文档时可能由于网络波动等原因无法下载或下载错误,付费完成后未能成功下载的用户请联系客服处理。