返回
Bayesian Information Recovery from CNN for Probabilistic Inference

Bayesian Information Recovery from CNN for Probabilistic Inference

ID：40363211

大小：5.24 MB

页数：8页

时间：2019-08-01

Bayesian Information Recovery from CNN for Probabilistic Inference_第1页
Bayesian Information Recovery from CNN for Probabilistic Inference_第2页
Bayesian Information Recovery from CNN for Probabilistic Inference_第3页
Bayesian Information Recovery from CNN for Probabilistic Inference_第4页
Bayesian Information Recovery from CNN for Probabilistic Inference_第5页
资源描述：

《Bayesian Information Recovery from CNN for Probabilistic Inference》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、2018IEEE/RSJInternationalConferenceonIntelligentRobotsandSystems(IROS)Madrid,Spain,October1-5,2018BayesianInformationRecoveryfromCNNforProbabilisticInferenceDmitryKopitkovandVadimIndelmanAbstract—Typicalinferenceapproachesthatworkwithhigh-dimensionalvisu

2、almeasurementsusehand-engineeredimagefeatures(e.g.SIFT)thatrequirecombinatorialdataassociation,orpredictonlyhiddenstatemeanwithoutconsideringitsuncertaintyandmulti-modalityaspects.WedevelopanovelapproachtoinfersystemhiddenstatefromvisualobservationsviaCN

3、NfeatureswhichareoutputsofaCNNclassifier.Tothatend,atpre-deploymentstageweuseneuralnetworksto(a)learnagenerativeviewpoint-dependentmodelofCNNfeaturesgiventherobotposeandapproximatethismodelbyaspatially-varyingGaussiandistribution.Further,atdeploymentthism

4、odelisutilizedwithinaBayesianframeworkforproba-bilisticinference,consideringarobotlocalizationproblem.Ourmethoddoesnotinvolvedataassociationandprovidesuncertaintycovarianceofthefinalestimation.Moreover,weshowempiricallythattheCNNfeaturelikelihoodisunimoda

5、lwhichsimplifiestheinferencetask.WetestourmethodinasimulatedUnrealEngineenvironment,wherewesucceedtoretrievehigh-levelstateinformationfromCNNfeaturesandproducetrajectoryestimationwithhighaccuracy.Additionally,weanalyzerobustnessofourapproachtodifferentlig

6、htconditions.I.INTRODUCTIONInferringasystemstatefrommultiplemeasurements,pos-siblycapturedbydifferentsensors,isafundamentalproblem(b)Fig.1:Approachoverview.InthispaperweuseCNNfeaturesforrobot’sstateinrobotics.Bayesianinferenceforsystemidentificationisinfe

7、rencewithinaBayesianframework.Animagecapturedfromrobotposexioneofthemainbuildingblocksonwhichmodernreal-ispassedtoaCNNclassifierwhichproducesafeaturesvectorfithatrepresentstheimage.(a)Duringthepre-deploymentstagewelearnspatially-varyingCNNworldroboticappl

8、icationsrely,suchasautonomousnavi-probabilitylikelihoodP(fijxi)approximatedbyN((xi);(xi)).Twoneuralgationandsimultaneouslocalizationandmapping(SLAM).networksproduceviewpoint-dependentmeanandcovariancefunctionsoffigivenxi

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。