서브메뉴
검색
Tips and Tricks for Building Controllable Artificial Intelligence
Tips and Tricks for Building Controllable Artificial Intelligence
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20260202103557
- ISBN
- 9798288862915
- DDC
- 004
- 저자명
- Mu, Norman.
- 서명/저자
- Tips and Tricks for Building Controllable Artificial Intelligence
- 발행사항
- [Sl] : University of California, Berkeley, 2025
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2025
- 형태사항
- 162 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 87-01, Section: B.
- 주기사항
- Advisor: Wagner, David A.
- 학위논문주기
- Thesis (Ph.D.)--University of California, Berkeley, 2025.
- 초록/해제
- 요약The end-to-end optimization of neural networks has led to tremendous advances in AI which are poised to disrupt many aspects of our lives. Congruently, an end-to-end engineering approach can help to ensure that these AI systems remain under the control of their users and developers. Design goals and threat models inform benchmarks and metrics, which inform training objectives and data, which inform neural architectures and algorithms. I discuss a variety of practical methods and considerations for this approach, including the benefits of multi-modality, assessing rule-based behaviors, challenges in securing broadly capable models, building robust control channels and safeguards, and more.
- 일반주제명
- Computer science
- 일반주제명
- Engineering
- 일반주제명
- Information technology
- 키워드
- Neural networks
- 키워드
- AI systems
- 기타저자
- University of California, Berkeley Computer Science
- 기본자료저록
- Dissertations Abstracts International. 87-01B.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008260126s2025 us c eng d■001000017357765
■00520260202103557
■006m o d
■007cr#unu||||||||
■020 ▼a9798288862915
■035 ▼a(MiAaPQ)AAI32042124
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a004
■1001 ▼aMu, Norman.
■24510▼aTips and Tricks for Building Controllable Artificial Intelligence
■260 ▼a[Sl]▼bUniversity of California, Berkeley▼c2025
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2025
■300 ▼a162 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 87-01, Section: B.
■500 ▼aAdvisor: Wagner, David A.
■5021 ▼aThesis (Ph.D.)--University of California, Berkeley, 2025.
■520 ▼aThe end-to-end optimization of neural networks has led to tremendous advances in AI which are poised to disrupt many aspects of our lives. Congruently, an end-to-end engineering approach can help to ensure that these AI systems remain under the control of their users and developers. Design goals and threat models inform benchmarks and metrics, which inform training objectives and data, which inform neural architectures and algorithms. I discuss a variety of practical methods and considerations for this approach, including the benefits of multi-modality, assessing rule-based behaviors, challenges in securing broadly capable models, building robust control channels and safeguards, and more.
■590 ▼aSchool code: 0028.
■650 4▼aComputer science
■650 4▼aEngineering
■650 4▼aInformation technology
■653 ▼aNeural networks
■653 ▼aAI systems
■653 ▼aRobust control channels
■653 ▼aEnd-to-end optimization
■690 ▼a0984
■690 ▼a0489
■690 ▼a0800
■690 ▼a0537
■71020▼aUniversity of California, Berkeley▼bComputer Science.
■7730 ▼tDissertations Abstracts International▼g87-01B.
■790 ▼a0028
■791 ▼aPh.D.
■792 ▼a2025
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17357765▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.
