毕业论文外文翻译-Caffe:快速特征嵌入的卷积结构

毕业论文外文翻译-Caffe:快速特征嵌入的卷积结构

ID:238908

大小:60.00 KB

页数:12页

时间:2017-07-12

毕业论文外文翻译-Caffe:快速特征嵌入的卷积结构_第1页
毕业论文外文翻译-Caffe:快速特征嵌入的卷积结构_第2页
毕业论文外文翻译-Caffe:快速特征嵌入的卷积结构_第3页
毕业论文外文翻译-Caffe:快速特征嵌入的卷积结构_第4页
毕业论文外文翻译-Caffe:快速特征嵌入的卷积结构_第5页
资源描述:

《毕业论文外文翻译-Caffe:快速特征嵌入的卷积结构》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、英文资料翻译Caffe:ConvolutionalArchitectureforFastFeatureEmbeddingABSTRACTCaffeprovidesmultimediascientistsandpractitionerswithacleanandmodiableframeworkforstate-of-the-artdeeplearningalgorithmsandacollectionofreferencemodels.TheframeworkisaBSD-licensedC++librarywithPythona

2、ndMATLABbindingsfortraininganddeployinggeneral-purposeconvolutionalneuralnetworksandotherdeepmodelsecientlyoncommodityarchitectures.Caetsindustryandinternet-scalemedianeedsbyCUDAGPUcomputation,processingover40millionimagesadayonasingleK40orTitanGPU(2.5msperimage).Byse

3、paratingmodelrepresentationfromactualimplementation,Caeallowsexperimentationandseamlessswitchingamongplatformsforeaseofdevelopmentanddeploymentfromprototypingma-chinestocloudenvironments.CaeismaintainedanddevelopedbytheBerkeleyVisionandLearningCenter(BVLC)withthehelpo

4、fanactivecommunityofcontributorsonGitHub.Itpowerson-goingresearchprojects,large-scaleindustrialapplications,andstartupprototypesinvision,speech,andmultimedia.CategoriesandSubjectDescriptorsI.5.1[PatternRecognition]:[Applications{Computervi-sion];D.2.2[SoftwareEngineer

5、ing]:[DesignToolsandTechniques{Softwarelibraries];I.5.1[PatternRecognition]:[Models{NeuralNets]GeneralTermsAlgorithms,Design,ExperimentationKeywordsOpenSource,ComputerVision,NeuralNetworks,ParallelComputation,MachineLearningCorrespondingAuthors.TheworkwasdonewhileYang

6、qingJiawasagraduatestudentatBerkeley.HeiscurrentlyaresearchscientistatGoogle,1600AmphitheaterPkwy,MountainView,CA94043.1.INTRODUCTIONAkeyprobleminmultimediadataanalysisisdiscoveryofeectiverepresentationsforsensoryinputs

7、images,sound-waves,haptics,etc.Whileperformanceo

8、fconventional,handcraftedfeatureshasplateauedinrecentyears,newdevelopmentsindeepcompositionalarchitectureshavekeptperformancelevelsrising[8].Deepmodelshaveoutperformedhand-engineeredfeaturerepresentationsinmanydo-mains,andmadelearningpossibleindomainswhereengineeredfe

9、atureswerelackingentirely.Weareparticularlymotivatedbylarge-scalevisualrecognition,whereaspecictypeofdeeparchitecturehasachi

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。