DC Field | Value | Language |
---|---|---|
dc.contributor.advisor | Rhu, Minsoo | - |
dc.contributor.advisor | 유민수 | - |
dc.contributor.author | Park, Beomsik | - |
dc.date.accessioned | 2023-06-26T19:34:20Z | - |
dc.date.available | 2023-06-26T19:34:20Z | - |
dc.date.issued | 2023 | - |
dc.identifier.uri | http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=1032937&flag=dissertation | en_US |
dc.identifier.uri | http://hdl.handle.net/10203/309961 | - |
dc.description | 학위논문(석사) - 한국과학기술원 : 전기및전자공학부, 2023.2,[iii, 24 p. :] | - |
dc.description.abstract | This paper proposed a new high-performant differentially private machine learning framework. It reduces the overall memory usage and increases training throughput while providing a mathematically same result. The proposed framework consisted of two major components, which are example-wise weight gradients computation and adaptive clipping. By implementing an end-to-end DP-SGD framework which utilizes these components, it is shown that a new framework can reduces the memory usage and increases training throughput. | - |
dc.language | eng | - |
dc.publisher | 한국과학기술원 | - |
dc.subject | 컴퓨터 아키텍처▼a머신 러닝▼a차등 정보보호 | - |
dc.subject | Differential privacy▼aMachine learning▼aComputer architecture | - |
dc.title | (A) fast and memory-efficient software framework for differentially private machine learning | - |
dc.title.alternative | 차등 정보보호 머신 러닝을 위한 메모리-효율적인 고속 소프트웨어 프레임워크 설계 | - |
dc.type | Thesis(Master) | - |
dc.identifier.CNRN | 325007 | - |
dc.description.department | 한국과학기술원 :전기및전자공학부, | - |
dc.contributor.alternativeauthor | 박범식 | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.