- 1、本文档共6页,可阅读全部内容。
- 2、原创力文档(book118)网站文档一经付费(服务费),不意味着购买了该文档的版权,仅供个人/单位学习、研究之用,不得用于商业用途,未经授权,严禁复制、发行、汇编、翻译或者网络传播等,侵权必究。
- 3、本站所有内容均由合作方或网友上传,本站不对文档的完整性、权威性及其观点立场正确性做任何保证或承诺!文档内容仅供研究参考,付费前请自行鉴别。如您付费,意味着您自己接受本站规则且自行承担风险,本站不退款、不进行额外附加服务;查看《如何避免下载的几个坑》。如果您已付费下载过本站文档,您可以点击 这里二次下载。
- 4、如文档侵犯商业秘密、侵犯著作权、侵犯人身权等,请点击“版权申诉”(推荐),也可以打举报电话:400-050-0827(电话支持时间:9:00-18:30)。
查看更多
深度学习的昨天、天和明天
ISSN
计算机研究与发展 1000—1%O/CN11—1777/TP
深度学习的昨天、今天和明天
余 凯 贾 磊 陈雨强 徐 伟
(百度北京 100085)
(yukai@baidu.corn)
Tomorrow
DeepLearning:Yesterday,Today,and
Yu Lei,Chen XuWei
Kai,Jia Yuqiang,and
100085)
(Baidu,Beijing
Machine isan areaofartificial Success
Abstract learningimportant intelligence.Since1980s,huge
hasbeenachievedintermsof newmachine
2006,a
algorithms,theory,andapplications.From
been intheresearch has
learningparadigm,nameddeeplearning,haspopular community,and
becomea waveof trendfor dataandartificial
huge technology big intelligence.Deeplearning
simulatesthehierarchicalstructureofhuman datafromlowerlevelto level,
brain,processing higher
and moreandmoresemantic recent
graduallycomposing concepts.In
a
IBM,andBaiduhaveinvestedlotofresourcesintotheRDofdeeplearning,makingsignificant
on online
progressesspeechrecognition,imageunderstanding,natural
languageprocessing,and
terms
ofthecontributiontoreal—world is themost
advertising.In applications,deeplearningperhaps
successful madethemachine in 10
thelast thisarticle,we
progressby learningcommunity years.In
will
文档评论(0)