서브메뉴
검색
Incidental Supervision for Natural Language Understanding- [electronic resource]
Incidental Supervision for Natural Language Understanding- [electronic resource]
상세정보
- 자료유형
- 학위논문파일 국외
- 최종처리일시
- 20240214100123
- ISBN
- 9798379751050
- DDC
- 004
- 저자명
- He, Hangfeng.
- 서명/저자
- Incidental Supervision for Natural Language Understanding - [electronic resource]
- 발행사항
- [S.l.]: : University of Pennsylvania., 2023
- 발행사항
- Ann Arbor : : ProQuest Dissertations & Theses,, 2023
- 형태사항
- 1 online resource(175 p.)
- 주기사항
- Source: Dissertations Abstracts International, Volume: 84-12, Section: A.
- 주기사항
- Advisor: Roth, Dan.
- 학위논문주기
- Thesis (Ph.D.)--University of Pennsylvania, 2023.
- 사용제한주기
- This item must not be sold to any third party vendors.
- 초록/해제
- 요약Acquiring human annotations for natural language understanding (NLU) tasks is often labor-intensive and demands significant domain-specific expertise, making it crucial to obtain supervision from indirect signals to improve target task performance. This dissertation presents a novel approach for enhancing NLU by harnessing on the power of incidental supervision signals, which are present in the data and the environment, regardless of the specific tasks being considered.The primary aim of this research is to deepen our understanding of incidental supervision signals and to develop efficient algorithms for their acquisition, selection, and usage in NLU tasks. This problem presents numerous challenges, including the intricate nature of natural language and the inherent disparities between incidental supervision signals and target tasks. To tackle these challenges, this dissertation employs a multifaceted approach. First, we demonstrate the feasibility of utilizing cost-effective signals to enhance various target tasks. Specifically, we retrieve signals from sentence-level question-answer pairs to help NLU tasks via two types of sentence encoding approaches, depending on whether the target task involves single- or multi-sentence input.Second, we introduce a unified informativeness measure to quantitatively assess the effectiveness of diverse incidental supervision signals for a given target task. This approach offers a promising way to determine, ahead of learning, which supervision signals would be beneficial.Finally, we present a suite of efficient algorithms for exploiting distinct types of incidental supervision signals, including a weighting strategy to enhance sample efficiency in cross-task learning and a post-processing technique for faithful large language model inference with external knowledge.
- 일반주제명
- Computer science.
- 일반주제명
- Information technology.
- 일반주제명
- Information science.
- 키워드
- Challenges
- 키워드
- Learning
- 키워드
- Target task
- 기타저자
- University of Pennsylvania Computer and Information Science
- 기본자료저록
- Dissertations Abstracts International. 84-12A.
- 기본자료저록
- Dissertation Abstract International
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008240612s2023 us |||||||||||||||c||eng d■001000016931821
■00520240214100123
■006m o d
■007cr#unu||||||||
■020 ▼a9798379751050
■035 ▼a(MiAaPQ)AAI30424791
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a004
■1001 ▼aHe, Hangfeng.
■24510▼aIncidental Supervision for Natural Language Understanding▼h[electronic resource]
■260 ▼a[S.l.]:▼bUniversity of Pennsylvania. ▼c2023
■260 1▼aAnn Arbor :▼bProQuest Dissertations & Theses, ▼c2023
■300 ▼a1 online resource(175 p.)
■500 ▼aSource: Dissertations Abstracts International, Volume: 84-12, Section: A.
■500 ▼aAdvisor: Roth, Dan.
■5021 ▼aThesis (Ph.D.)--University of Pennsylvania, 2023.
■506 ▼aThis item must not be sold to any third party vendors.
■520 ▼aAcquiring human annotations for natural language understanding (NLU) tasks is often labor-intensive and demands significant domain-specific expertise, making it crucial to obtain supervision from indirect signals to improve target task performance. This dissertation presents a novel approach for enhancing NLU by harnessing on the power of incidental supervision signals, which are present in the data and the environment, regardless of the specific tasks being considered.The primary aim of this research is to deepen our understanding of incidental supervision signals and to develop efficient algorithms for their acquisition, selection, and usage in NLU tasks. This problem presents numerous challenges, including the intricate nature of natural language and the inherent disparities between incidental supervision signals and target tasks. To tackle these challenges, this dissertation employs a multifaceted approach. First, we demonstrate the feasibility of utilizing cost-effective signals to enhance various target tasks. Specifically, we retrieve signals from sentence-level question-answer pairs to help NLU tasks via two types of sentence encoding approaches, depending on whether the target task involves single- or multi-sentence input.Second, we introduce a unified informativeness measure to quantitatively assess the effectiveness of diverse incidental supervision signals for a given target task. This approach offers a promising way to determine, ahead of learning, which supervision signals would be beneficial.Finally, we present a suite of efficient algorithms for exploiting distinct types of incidental supervision signals, including a weighting strategy to enhance sample efficiency in cross-task learning and a post-processing technique for faithful large language model inference with external knowledge.
■590 ▼aSchool code: 0175.
■650 4▼aComputer science.
■650 4▼aInformation technology.
■650 4▼aInformation science.
■653 ▼aNatural language understanding
■653 ▼aSupervision signals
■653 ▼aHuman annotations
■653 ▼aChallenges
■653 ▼aLearning
■653 ▼aTarget task
■690 ▼a0984
■690 ▼a0489
■690 ▼a0723
■71020▼aUniversity of Pennsylvania▼bComputer and Information Science.
■7730 ▼tDissertations Abstracts International▼g84-12A.
■773 ▼tDissertation Abstract International
■790 ▼a0175
■791 ▼aPh.D.
■792 ▼a2023
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T16931821▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.
■980 ▼a202402▼f2024
![Incidental Supervision for Natural Language Understanding - [electronic resource]](/Users/Baul/Images/book.png)

