Energy-Efficient Context Classification With英文电子资料

Energy-Efficient Context Classification With英文电子资料

ID:39326428

大小:432.83 KB

页数:12页

时间:2019-06-30

Energy-Efficient Context Classification With英文电子资料_第1页
Energy-Efficient Context Classification With英文电子资料_第2页
Energy-Efficient Context Classification With英文电子资料_第3页
Energy-Efficient Context Classification With英文电子资料_第4页
Energy-Efficient Context Classification With英文电子资料_第5页
资源描述:

《Energy-Efficient Context Classification With英文电子资料》由会员上传分享,免费在线阅读,更多相关内容在行业资料-天天文库

1、IEEETRANSACTIONSONBIOMEDICALCIRCUITSANDSYSTEMS,VOL.6,NO.2,APRIL2012167Energy-EfficientContextClassificationWithDynamicSensorControlLawrenceK.Au,StudentMember,IEEE,AlexA.T.Bui,Member,IEEE,MaximA.Batalin,Member,IEEE,andWilliamJ.Kaiser,SeniorMember,IEEEAbstract—Energyefficiencyhasbeenalongstandin

2、gdesignprocessingalgorithmsmustprovideadequateinformationthatchallengeforwearablesensorsystems.Itisespeciallycrucialdescribethepatients,butthewearablesystemmustbelight-incontinuoussubjectstatemonitoringduetotheongoingneedweightenoughtobeworninreallife.forcompactsizesandbettersensors.Thispap

3、erpresentsanAmongallthedesignrequirements,energyefficiencyhasenergy-efficientclassificationalgorithm,basedonpartiallyob-servableMarkovdecisionprocess(POMDP).Ineverytimestep,beenamajordesignconsiderationofwearableplatforms[10],POMDPdynamicallyselectssensorsforclassificationviaasensor[11].Battery

4、lifeisoftenseverelylimitedinwearableappli-selectionpolicy.Thesensorselectionproblemisformalizedascations,yetconstantclassificationandwirelesssensinginduceanoptimizationproblem,wheretheobjectiveistominimizesignificantenergyconsumption.Manyexistingapplicationsdomisclassificationcostgivensomeener

5、gybudget.Statetransitionsnotaddressthisproblemdirectlybecausethelengthoftheirex-aremodeledasahiddenMarkovmodel(HMM),andthecorre-spondingsensorselectionpolicyisrepresentedusingafinite-stateperimentsandpilotstudiesarerelativelyshortandthedepictedcontroller(FSC).Toevaluatethisframework,sensorda

6、taweresituationsareoftencontrolled.Suchscenariosmaynotbeap-collectedfrommultiplesubjectsintheirfree-livingconditions.plicableinpractice:batteryenergycanbecompletelydepletedRelativeaccuraciesandenergyreductionsfromtheproposedwithinhoursifnotmanagedproperly[12],andrechargingismethodarecompare

7、dagainstnaïveBayes(always-on)andsimpleoftenunavailablewhilethesystemsareinuse.Consequently,randomstrategiestovalidatetherelativeperformanceoftheal-gorithm.Whentheobjectiveistomaintainthesameclassificationmoreenergy-efficienttechniquesareneededtomanagebatte

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。