서브메뉴
검색
Efficient Algorithms for Modern Machine Learning Optimization
Efficient Algorithms for Modern Machine Learning Optimization
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20260202103619
- ISBN
- 9798315793328
- DDC
- 519
- 저자명
- Ding, Lisang.
- 서명/저자
- Efficient Algorithms for Modern Machine Learning Optimization
- 발행사항
- [Sl] : University of California, Los Angeles, 2025
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2025
- 형태사항
- 135 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 86-12, Section: B.
- 주기사항
- Advisor: Osher, Stanley J.;Yin, Wotao.
- 학위논문주기
- Thesis (Ph.D.)--University of California, Los Angeles, 2025.
- 초록/해제
- 요약The training of machine learning models is a central topic in modern computational research and is often formulated as an optimization problem. However, the optimization landscape in machine learning presents new and complex challenges. This dissertation addresses two related optimization problems: decentralized optimization, which arises in distributed training settings, and sum-of-minimum optimization, which emerges in mixed-model training.For the decentralized optimization problem, we revisit and construct a communication-optimal exact consensus scheme. This scheme is then judiciously integrated into the decentralized stochastic gradient descent algorithm. The proposed decentralized algorithm is scalable to any number of computing nodes and achieves state-of-the-art performance, which requires a transient iteration complexity of O(n3) and a communication overhead of one.For the sum-of-minimum optimization problem, we identify a novel connection to generalized clustering problems. Leveraging this insight, we develop a two-phase algorithm. In the initialization phase, we generalize the k-means++ clustering method; in the iteration phase, we apply a variant of the Lloyd algorithm. Theoretically, a tight initialization error bound and a convergence rate are provided.For both problems, extensive numerical experiments are provided to illustrate the empirical performance of the proposed algorithms.
- 일반주제명
- Applied mathematics
- 일반주제명
- Mathematics
- 일반주제명
- Computational physics
- 키워드
- Machine learning
- 키워드
- Lloyd algorithm
- 기타저자
- University of California, Los Angeles Mathematics 0540
- 기본자료저록
- Dissertations Abstracts International. 86-12B.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008260126s2025 us c eng d■001000017357930
■00520260202103619
■006m o d
■007cr#unu||||||||
■020 ▼a9798315793328
■035 ▼a(MiAaPQ)AAI32045196
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a519
■1001 ▼aDing, Lisang.
■24510▼aEfficient Algorithms for Modern Machine Learning Optimization
■260 ▼a[Sl]▼bUniversity of California, Los Angeles▼c2025
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2025
■300 ▼a135 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 86-12, Section: B.
■500 ▼aAdvisor: Osher, Stanley J.;Yin, Wotao.
■5021 ▼aThesis (Ph.D.)--University of California, Los Angeles, 2025.
■520 ▼aThe training of machine learning models is a central topic in modern computational research and is often formulated as an optimization problem. However, the optimization landscape in machine learning presents new and complex challenges. This dissertation addresses two related optimization problems: decentralized optimization, which arises in distributed training settings, and sum-of-minimum optimization, which emerges in mixed-model training.For the decentralized optimization problem, we revisit and construct a communication-optimal exact consensus scheme. This scheme is then judiciously integrated into the decentralized stochastic gradient descent algorithm. The proposed decentralized algorithm is scalable to any number of computing nodes and achieves state-of-the-art performance, which requires a transient iteration complexity of O(n3) and a communication overhead of one.For the sum-of-minimum optimization problem, we identify a novel connection to generalized clustering problems. Leveraging this insight, we develop a two-phase algorithm. In the initialization phase, we generalize the k-means++ clustering method; in the iteration phase, we apply a variant of the Lloyd algorithm. Theoretically, a tight initialization error bound and a convergence rate are provided.For both problems, extensive numerical experiments are provided to illustrate the empirical performance of the proposed algorithms.
■590 ▼aSchool code: 0031.
■650 4▼aApplied mathematics
■650 4▼aMathematics
■650 4▼aComputational physics
■653 ▼aMachine learning
■653 ▼aOptimization problem
■653 ▼aLloyd algorithm
■653 ▼aStochastic gradient
■690 ▼a0364
■690 ▼a0405
■690 ▼a0216
■71020▼aUniversity of California, Los Angeles▼bMathematics 0540.
■7730 ▼tDissertations Abstracts International▼g86-12B.
■790 ▼a0031
■791 ▼aPh.D.
■792 ▼a2025
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17357930▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.
