서브메뉴
검색
Transitional Information Extraction
Transitional Information Extraction
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20260202105601
- ISBN
- 9798265404947
- DDC
- 000
- 저자명
- Benkert, Ryan.
- 서명/저자
- Transitional Information Extraction
- 발행사항
- [Sl] : Georgia Institute of Technology, 2024
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2024
- 형태사항
- 155 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 87-05, Section: B.
- 주기사항
- Advisor: AlRegib, Ghassan.
- 학위논문주기
- Thesis (Ph.D.)--Georgia Institute of Technology, 2024.
- 초록/해제
- 요약The practical deployment of neural networks is contingent on their ability to extract information beyond a predictive output. Unfortunately, neural networks do not naturally implement this property and are optimized to compress information unrelated to the target application. For this purpose, extracting information from neural networks is frequently coupled with optimization constraints or with utilizing a collection of neural networks that require several iterative forward passes. While each of these paradigms show promising results, each comes with a set of drawbacks that limit their deployment in the real world: constraints can result in a performance degradation on the application while iterative approaches require several forward passes.In this thesis, we address these limitations by realizing iterative information extraction as a sub-category of a broader family of paradigms transitional information extraction. A key observation we make in this thesis is that the source representations are generalizable beyond several forward passes and can be implemented within a single neural network. Our approach is grounded in a detailed analysis of existing paradigms where we identify causes of performance decline for application constraints, as well as implication of limited data on the information extraction capabilities of the network. Our study results in a suite of algorithms that implement transitional information extraction. We study transitional information extraction within the context of uncertainty estimation. Here, we develop TULIP which leverages intermediate representations to preserve information as data points traverse the network. We further investigate active learning where we consider using prediction switches to extract different information from the neural network. We extract information related to robustness with GauSS, performance regression with Rose, and spatial regions of interest with ATLAS. All of our algorithms are supported with extensive experiments on established benchmarks within the field of active learning and uncertainty estimation.
- 일반주제명
- Toads
- 일반주제명
- Deep learning
- 일반주제명
- Entropy
- 일반주제명
- Neural networks
- 일반주제명
- Benchmarks
- 기본자료저록
- Dissertations Abstracts International. 87-05B.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008260126s2024 us c eng d■001000017360648
■00520260202105601
■006m o d
■007cr#unu||||||||
■020 ▼a9798265404947
■035 ▼a(MiAaPQ)AAI32315974
■035 ▼a(MiAaPQ)GeorgiaTech75284
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a000
■1001 ▼aBenkert, Ryan.
■24510▼aTransitional Information Extraction
■260 ▼a[Sl]▼bGeorgia Institute of Technology▼c2024
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2024
■300 ▼a155 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 87-05, Section: B.
■500 ▼aAdvisor: AlRegib, Ghassan.
■5021 ▼aThesis (Ph.D.)--Georgia Institute of Technology, 2024.
■520 ▼aThe practical deployment of neural networks is contingent on their ability to extract information beyond a predictive output. Unfortunately, neural networks do not naturally implement this property and are optimized to compress information unrelated to the target application. For this purpose, extracting information from neural networks is frequently coupled with optimization constraints or with utilizing a collection of neural networks that require several iterative forward passes. While each of these paradigms show promising results, each comes with a set of drawbacks that limit their deployment in the real world: constraints can result in a performance degradation on the application while iterative approaches require several forward passes.In this thesis, we address these limitations by realizing iterative information extraction as a sub-category of a broader family of paradigms transitional information extraction. A key observation we make in this thesis is that the source representations are generalizable beyond several forward passes and can be implemented within a single neural network. Our approach is grounded in a detailed analysis of existing paradigms where we identify causes of performance decline for application constraints, as well as implication of limited data on the information extraction capabilities of the network. Our study results in a suite of algorithms that implement transitional information extraction. We study transitional information extraction within the context of uncertainty estimation. Here, we develop TULIP which leverages intermediate representations to preserve information as data points traverse the network. We further investigate active learning where we consider using prediction switches to extract different information from the neural network. We extract information related to robustness with GauSS, performance regression with Rose, and spatial regions of interest with ATLAS. All of our algorithms are supported with extensive experiments on established benchmarks within the field of active learning and uncertainty estimation.
■590 ▼aSchool code: 0078.
■650 4▼aToads
■650 4▼aDeep learning
■650 4▼aEntropy
■650 4▼aNeural networks
■650 4▼aBenchmarks
■690 ▼a0800
■71020▼aGeorgia Institute of Technology.
■7730 ▼tDissertations Abstracts International▼g87-05B.
■790 ▼a0078
■791 ▼aPh.D.
■792 ▼a2024
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17360648▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.


