서브메뉴
검색
Compute-Constrained Continual Learning: Foundations and Algorithms
Compute-Constrained Continual Learning: Foundations and Algorithms
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20260202105056
- ISBN
- 9798288816277
- DDC
- 658
- 저자명
- Kumar, Saurabh.
- 서명/저자
- Compute-Constrained Continual Learning: Foundations and Algorithms
- 발행사항
- [Sl] : Stanford University, 2025
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2025
- 형태사항
- 141 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 87-01, Section: B.
- 주기사항
- Advisor: Finn, Chelsea;Van Roy, Benjamin.
- 학위논문주기
- Thesis (Ph.D.)--Stanford University, 2025.
- 초록/해제
- 요약Continual learning remains a long-standing challenge of machine learning. Success requires continuously ingesting new knowledge while retaining old knowledge that remains useful. In this thesis, we introduce a coherent objective for continual learning based on maximizing infinite-horizon average reward under a per-timestep computational constraint. This framing allows us to systematically reason about the design and evaluation of continual learning agents, moving beyond ad hoc metrics like accuracy retention or plasticity alone. Part I of the thesis develops foundational tools and perspectives, including an information-theoretic treatment of agent state, a quantification of information capacity, and an exploration of the stability-plasticity trade-off in continual learning. Part II presents new algorithms: a regenerative regularization method to combat plasticity loss in neural networks, Conformal Sympow--a transformer-based model that enables efficient long-context inference via learned gating and data-dependent rotations, and a diversity-driven reinforcement learning approach that enables few-shot robustness to environment perturbations. Together, these contributions help ground continual learning as a principled and tractable subfield of machine learning, bridging theory and practice.
- 일반주제명
- Behavior
- 일반주제명
- Probability
- 일반주제명
- Electricity
- 일반주제명
- Large language models
- 일반주제명
- Neural networks
- 일반주제명
- Initiatives
- 일반주제명
- Computer engineering
- 키워드
- Machine learning
- 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■001000017359292
■00520260202105056
■006m o d
■007cr#unu||||||||
■020 ▼a9798288816277
■035 ▼a(MiAaPQ)AAI32201037
■035 ▼a(MiAaPQ)Stanfordzp439vy6712
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a658
■1001 ▼aKumar, Saurabh.
■24510▼aCompute-Constrained Continual Learning: Foundations and Algorithms
■260 ▼a[Sl]▼bStanford University▼c2025
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2025
■300 ▼a141 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 87-01, Section: B.
■500 ▼aAdvisor: Finn, Chelsea;Van Roy, Benjamin.
■5021 ▼aThesis (Ph.D.)--Stanford University, 2025.
■520 ▼aContinual learning remains a long-standing challenge of machine learning. Success requires continuously ingesting new knowledge while retaining old knowledge that remains useful. In this thesis, we introduce a coherent objective for continual learning based on maximizing infinite-horizon average reward under a per-timestep computational constraint. This framing allows us to systematically reason about the design and evaluation of continual learning agents, moving beyond ad hoc metrics like accuracy retention or plasticity alone. Part I of the thesis develops foundational tools and perspectives, including an information-theoretic treatment of agent state, a quantification of information capacity, and an exploration of the stability-plasticity trade-off in continual learning. Part II presents new algorithms: a regenerative regularization method to combat plasticity loss in neural networks, Conformal Sympow--a transformer-based model that enables efficient long-context inference via learned gating and data-dependent rotations, and a diversity-driven reinforcement learning approach that enables few-shot robustness to environment perturbations. Together, these contributions help ground continual learning as a principled and tractable subfield of machine learning, bridging theory and practice.
■590 ▼aSchool code: 0212.
■650 4▼aBehavior
■650 4▼aProbability
■650 4▼aElectricity
■650 4▼aLarge language models
■650 4▼aNeural networks
■650 4▼aInitiatives
■650 4▼aComputer engineering
■653 ▼aMachine learning
■653 ▼aContinual learning
■653 ▼aAccuracy retention
■690 ▼a0464
■690 ▼a0800
■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=T17359292▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.
