Publications

Detailed Information

Layer-wise Progressive Knowledge Distillation : 지식 증류를위한 다단계 교사

DC Field Value Language
dc.contributor.advisorLee, Kyoung Mu-
dc.contributor.authorMohammad Amin Shabani-
dc.date.accessioned2019-10-18T15:42:49Z-
dc.date.available2019-10-18T15:42:49Z-
dc.date.issued2019-08-
dc.identifier.other000000157560-
dc.identifier.urihttps://hdl.handle.net/10371/161065-
dc.identifier.urihttp://dcollection.snu.ac.kr/common/orgView/000000157560ko_KR
dc.description학위논문(석사)--서울대학교 대학원 :공과대학 전기·컴퓨터공학부,2019. 8. Lee, Kyoung Mu.-
dc.description.abstract지식 증류 (Knowledge Distillation, KD)는 교사로부터 학생 모델로 지식을 전 달하는 잘 알려진 방법입니다. 본 논문에서는 계층 적 진보적 교사 (Layer-wise Pro- gressive Teacher)를 도입하여 지식 증류를위한 새로운 틀을 제안하고자한다. 이와 관련하여 우리는 교사의 중간 계층에서 확률을 구함으로써 서로 다른 경도 수준에 서 부드러운 목표를 만드는 방법을 제안합니다. 우리의 방법은 교사와 학생 사이에 큰 차이가있어 학생이 교사를 모방하는 것을 더 어렵게하는 경우를 위해 특별히 고 안되었습니다. 우리는 또한 학생의 온도를 제거하고 교사의 온도를 유지하는 것이 좋습니다. 실험 결과는 기존의 증류법과 비교할 때 우리의 방법이 훨씬 더 우수한 결과를 얻음을 보여줍니다.-
dc.description.abstractKnowledge Distillation (KD) is a well-known method for transferring knowledge from a teacher to a student model. In this thesis, we propose a new framework for Knowledge Distillation by introducing a Layer-wise Progressive Teacher. In this regard, we propose a method to create soft targets in different levels of complexity by obtaining the probabilities from the intermediate layers of the teacher network. Our method is specially designed for the cases that there is a large gap between the teacher and the student which makes it harder for the student to mimic the teacher. In addition, we proposed focalized teacher as a method to train a better teacher for the student. The experimental results show that our method gets significantly better results in comparison with existing knowledge distillation methods.-
dc.description.tableofcontents1 Introduction 1
1.1 Background. 1
1.2 Motivation 3
1.3 ProposedMethod 4
1.4 Datasets. 5
2 Related Work 7
2.1 Theory of Transfer Learning 7
2.2 Applications. 8
3 Focalized Teacher 10
3.1 Overview 10
3.2 LabelCorrection 11
3.3 FocalizedTeacher 12
3.4 Experimental Results 13
4 Layer-wise Progressive Knowledge Distillation 16
4.1 BackgroundandNotations . 16
4.2 Layer-wiseKnowledgeDistillation. 17
4.3 ProgressiveTeacher. 19
4.4 Experimental Results 20
4.4.1 TemperatureAnalysis . 21
4.4.2 DistanceMetric. 23
4.4.3 Distilled Knowledge from an intermediate layer . . . . . . . . 24
4.4.4 ProgressiveTeacher 27
4.4.5 ComparisonwithotherKDmethods 29
5 Conclusion
5.1 SummaryoftheThesis . 32
5.2 FutureWorks 32
5.2.1 Progressive Teacher Assistant based Knowledge Distillation . 33
-
dc.language.isoeng-
dc.publisher서울대학교 대학원-
dc.subjectKnowledge Distillation-
dc.subjectTransfer Learning-
dc.subjectImage Classification-
dc.subject.ddc621.3-
dc.titleLayer-wise Progressive Knowledge Distillation-
dc.title.alternative지식 증류를위한 다단계 교사-
dc.typeThesis-
dc.typeDissertation-
dc.contributor.AlternativeAuthor모하마드-
dc.contributor.department공과대학 전기·컴퓨터공학부-
dc.description.degreeMaster-
dc.date.awarded2019-08-
dc.identifier.uciI804:11032-000000157560-
dc.identifier.holdings000000000040▲000000000041▲000000157560▲-
Appears in Collections:
Files in This Item:

Altmetrics

Item View & Download Count

  • mendeley

Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.

Share