DC Field | Value | Language |
---|---|---|
dc.contributor.advisor | Kim, Lee-Sup | - |
dc.contributor.advisor | 김이섭 | - |
dc.contributor.author | 신재강 | - |
dc.date.accessioned | 2022-04-27T19:31:00Z | - |
dc.date.available | 2022-04-27T19:31:00Z | - |
dc.date.issued | 2020 | - |
dc.identifier.uri | http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=986295&flag=dissertation | en_US |
dc.identifier.uri | http://hdl.handle.net/10203/295945 | - |
dc.description | 학위논문(석사) - 한국과학기술원 : 전기및전자공학부, 2020.2,[iv, 34 p. :] | - |
dc.description.abstract | Incremental learning is drawing attention to widen capabilities of device-AI. Previous works have researched to reduce numerous computations and memory accesses required for the training process of IL, but they could not show a noticeable improvement in the weight gradient computation (WGC) phase. Therefore, a selective weight update technique that searches for critical weights to be updated is pro-posed by applying the IL algorithm that training per-task binary masks. Also, a novel dataflow for the implementation of selective WGC is introduced on typical NPUs with minimum overheads. On average, the proposed system shows a 2.91x speed up and 2.48x energy efficiency in WGC without degrading training quality. | - |
dc.language | eng | - |
dc.publisher | 한국과학기술원 | - |
dc.subject | On-device AI▼aCNN▼aIncremnetal learning▼aSelective weight gradient computation▼aDNN accelerator | - |
dc.subject | 온 디바이스 인공지능▼a컨볼루셔널 뉴럴 네트워크▼a증분 학습▼a선택적 웨이트 그래디언트 연산▼a딥뉴럴넷 가속기 | - |
dc.title | (A) pragmatic approach to on-device incremental learning system with selective weight updates | - |
dc.title.alternative | 선택적 웨이트 업데이트를 통한 기기 내 증분 학습 시스템에 대한 실용적인 접근 | - |
dc.type | Thesis(Master) | - |
dc.identifier.CNRN | 325007 | - |
dc.description.department | 한국과학기술원 :전기및전자공학부, | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.