서브메뉴
검색
High Dimensionality in Modern Machine Learning: A Random Matrix Theory Perspective
High Dimensionality in Modern Machine Learning: A Random Matrix Theory Perspective
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20260202104849
- ISBN
- 9798288816611
- DDC
- 614.4
- 저자명
- Cheng, Chen.
- 서명/저자
- High Dimensionality in Modern Machine Learning: A Random Matrix Theory Perspective
- 발행사항
- [Sl] : Stanford University, 2025
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2025
- 형태사항
- 283 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 87-01, Section: B.
- 주기사항
- Advisor: Duchi, John;Montanari, Andrea.
- 학위논문주기
- Thesis (Ph.D.)--Stanford University, 2025.
- 초록/해제
- 요약This thesis studies curious phenomena arising from high dimensionality in modern machine learning datasets, with an emphasis on developing theoretical understandings through the lens of random matrix theory (RMT). High dimensional datasets are not foreign topics to statisticians, and conventional methods typically focus on sparsity and model selection, operating under the assumption that the underlying dimensionality of the data is low. However, modern complex and overparameterized machine learning models break this standard assumption. Moreover, intriguing phenomena such as memorization, double descent, and benign overfitting have emerged, raising new questions about the classical textbook predictions regarding the dichotomy between underfitting and overfitting. Random matrix theory has emerged as a valuable tool for analyzing these phenomena, while new setups and challenges simultaneously drive the development of innovative techniques within RMT. The thesis provides theoretical insights for three specific modern machine learning problems, harnessing the power of RMT as well as developing new technical tools in RMT.
- 일반주제명
- Pandemics
- 일반주제명
- Computer engineering
- 기타저자
- Stanford University.
- 기본자료저록
- Dissertations Abstracts International. 87-01B.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008260126s2025 us c eng d■001000017359206
■00520260202104849
■006m o d
■007cr#unu||||||||
■020 ▼a9798288816611
■035 ▼a(MiAaPQ)AAI32200941
■035 ▼a(MiAaPQ)Stanfordfq431rr3697
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a614.4
■1001 ▼aCheng, Chen.
■24510▼aHigh Dimensionality in Modern Machine Learning: A Random Matrix Theory Perspective
■260 ▼a[Sl]▼bStanford University▼c2025
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2025
■300 ▼a283 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 87-01, Section: B.
■500 ▼aAdvisor: Duchi, John;Montanari, Andrea.
■5021 ▼aThesis (Ph.D.)--Stanford University, 2025.
■520 ▼aThis thesis studies curious phenomena arising from high dimensionality in modern machine learning datasets, with an emphasis on developing theoretical understandings through the lens of random matrix theory (RMT). High dimensional datasets are not foreign topics to statisticians, and conventional methods typically focus on sparsity and model selection, operating under the assumption that the underlying dimensionality of the data is low. However, modern complex and overparameterized machine learning models break this standard assumption. Moreover, intriguing phenomena such as memorization, double descent, and benign overfitting have emerged, raising new questions about the classical textbook predictions regarding the dichotomy between underfitting and overfitting. Random matrix theory has emerged as a valuable tool for analyzing these phenomena, while new setups and challenges simultaneously drive the development of innovative techniques within RMT. The thesis provides theoretical insights for three specific modern machine learning problems, harnessing the power of RMT as well as developing new technical tools in RMT.
■590 ▼aSchool code: 0212.
■650 4▼aPandemics
■650 4▼aComputer engineering
■653 ▼aRandom matrix theory
■653 ▼aMachine learning datasets
■690 ▼a0800
■690 ▼a0464
■71020▼aStanford University.
■7730 ▼tDissertations Abstracts International▼g87-01B.
■790 ▼a0212
■791 ▼aPh.D.
■792 ▼a2025
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17359206▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.
