> \pJava Excel API v2.6 Ba==h\:#8X@"1Arial1Arial1Arial1Arial + ) , * `NDC,B4$titletitle[alternative]title[translated]author[google(nobreakline)] contributor[author(nobreakline)]"contributor[scopusid(nobreakline)]date[issued]relation[journaltitle]identifier[issn] citationsidentifier[major]subject(nobreakline) publisheridentifier[thesisdegree]contributor[advisor]relation[ispartofseries]relation[index(nobreakline)]typeabstractidentifier[govdoc]identifier[uri]identifier[doi(doi)]identifier[isbn]identifier[ismn]
identifier%ݲ )`X ttThe review of Deep learning2019Y ĬYtTŐYP YMasterȰMaster's Thesis%ݲ t`t L \ @ 1980Dt. X̹ Xܴƴ@ pt0, \T LବX \Ĭ\ xt ȩ . 췘 = tܴ, x07, LବX hخ 2010D0 x 8X tհE<\ %ݲt pȅ 0 ǈ. <\ t x t \ 8 %ݲ @ 0tX 8ݲ lЌ Uĳ| , 0| %ݲ \ l ,Ґĳ \ t踴 .
t\ Pׄ ʹ \x %ݲ lpx 5 ½ݹ(MLP), i1 ½ݹ(CNN), X ½ݹ(RNN)D |8 Xଐ \. п̹ DȲ|, ` ǔ 8 t| DT¬ ǔ )ĳ hخ Xଐ \. ȹɹ<\ )`X ȩ | hخ . ٳ|\ pt0 Ÿ D iX DPX. (Kaggle) XLѤ¸ %ݲ D t ǔ Ÿ X DP| t )`X ĳ hخ tଐ \. |8X XXǔ \x %ݲ )`D ttX, i Xt| ` D xXp, %ݲ Y i \ ĳl| ` ĳ] ճ p .;Although the theory of deep learning appeared in 1980s, it is from 2010s that deep learning took off as a solution of perceptual problems, thanks to breakthroughs in hardware, database and optimizer. Consequently deep learning has broken the records of traditional machine learning approaches in the fields such as image processing. This achievement led to active research and massive investment.
To keep up with this trend of data science society, this paper introduces commonly used structures of deep learning; multilayer perceptron, convolutional neural network, recurrent neural network. Some methods to mitigate overfitting are also written in this paper. Furthermore, the comparison of deep learning and gradient boosting, two approaches dominating Kaggle contests recently, is done by studying results of them on the same dataset at the end of this paper.khttp://dcollection.ewha.ac.kr/common/orgView/000000153528;
http://dspace.ewha.ac.kr/handle/2015.oak/248600;"@Ce|(!CPrs BKm~ AJlm)<3UBdKmlKm
"
dMbP?_*+%" ,,??U
!
"
#
l>@
Root EntryWorkbookSummaryInformation(DocumentSummaryInformation8