DC Field | Value | Language |
---|---|---|
dc.contributor.advisor | Oh, Alice | - |
dc.contributor.advisor | 오혜연 | - |
dc.contributor.author | Kim, Dongkwan | - |
dc.date.accessioned | 2021-05-11T19:35:27Z | - |
dc.date.available | 2021-05-11T19:35:27Z | - |
dc.date.issued | 2019 | - |
dc.identifier.uri | http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=876079&flag=dissertation | en_US |
dc.identifier.uri | http://hdl.handle.net/10203/283161 | - |
dc.description | 학위논문(석사) - 한국과학기술원 : 전산학부, 2019.8,[iv, 23 p. :] | - |
dc.description.abstract | Pruning units such as filters and neurons have been used to address the inefficiency of memory and computational resources in deep neural networks. Existing work mainly focuses on a single task with a static number of classes, but in practice, classes and tasks become obsolete. Using information about obsolescence, I propose a `Selective Forgetting' method that prunes units that are unnecessary to classes or tasks of interests. Furthermore, I suggest a novel neural architecture that is advantageous to selective forgetting by disentangling contributions of units across classes and tasks. I validate my approach in experiments about class forgetting on a single task model and task forgetting on continual learning models. Experimental results demonstrate that 1) my pruning method outperforms baselines in terms of mean and minimum performance of classes/tasks to be preserved and 2) my architecture makes units be disentangled and gains more benefit from selective forgetting than the entangled network. | - |
dc.language | eng | - |
dc.publisher | 한국과학기술원 | - |
dc.subject | Deep neural network pruning▼aefficient inference▼aclass and task obsolescence▼acontinual learning | - |
dc.subject | 심층 신경망 프루닝▼a효율적 추론▼a불용 클래스와 태스크▼a연속 학습 | - |
dc.title | Selective forgetting of classes and tasks for deep neural networks | - |
dc.title.alternative | 심층 신경망을 위한 클래스와 태스크의 선택적 망각 | - |
dc.type | Thesis(Master) | - |
dc.identifier.CNRN | 325007 | - |
dc.description.department | 한국과학기술원 :전산학부, | - |
dc.contributor.alternativeauthor | 김동관 | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.