본문

서브메뉴

Compositional Visual Learning and Reasoning in Humans and Machines
Compositional Visual Learning and Reasoning in Humans and Machines
Compositional Visual Learning and Reasoning in Humans and Machines

Detailed Information

자료유형  
 학위논문 서양
최종처리일시  
20250211152706
ISBN  
9798342709873
DDC  
153
저자명  
Zhou, Yanli.
서명/저자  
Compositional Visual Learning and Reasoning in Humans and Machines
발행사항  
[Sl] : New York University, 2024
발행사항  
Ann Arbor : ProQuest Dissertations & Theses, 2024
형태사항  
103 p
주기사항  
Source: Dissertations Abstracts International, Volume: 86-04, Section: B.
주기사항  
Advisor: Lake, Brenden M.
학위논문주기  
Thesis (Ph.D.)--New York University, 2024.
초록/해제  
요약Humans leverage compositionality to efficiently learn new visual concepts, understanding how familiar parts can be combined to form novel objects. Moreover, our conceptual knowledge of these novel objects is often enhanced by understanding how parts relate to their functions and how composing the parts relates to composing the functions. In contrast, popular computer vision models struggle to make the same types of inferences, requiring more data and generalizing less flexibly than people do.In this thesis, we explore two aspects of human visual learning and reasoning. We first study the unique human capability to learn visual concepts across a range of different types of visual composition, examining how people classify and generate a novel class of abstract visual concepts with rich relational structures. We also examine how functions are learned and represented by humans from limited visual observations and the ability to perform zero-shot function compositions in various interaction conditions. In all studies, we compare humans with computational models directly on the same set of learning tasks, shedding light on whether computational models, including Bayesian program induction models and meta-learning neural networks, can perform compositional visual concept and function learning in the complete and complex way humans do.
일반주제명  
Cognitive psychology
일반주제명  
Computer science
일반주제명  
Psychology
키워드  
Bayesian modeling
키워드  
Cognitive modeling
키워드  
Compositionality
키워드  
Concept learning
키워드  
Meta-learning
키워드  
Reasoning
기타저자  
New York University Center for Data Science
기본자료저록  
Dissertations Abstracts International. 86-04B.
전자적 위치 및 접속  
로그인 후 원문을 볼 수 있습니다.

MARC

 008250123s2024        us                              c    eng  d
■001000017163427
■00520250211152706
■006m          o    d                
■007cr#unu||||||||
■020    ▼a9798342709873
■035    ▼a(MiAaPQ)AAI31488377
■040    ▼aMiAaPQ▼cMiAaPQ
■0820  ▼a153
■1001  ▼aZhou,  Yanli.
■24510▼aCompositional  Visual  Learning  and  Reasoning  in  Humans  and  Machines
■260    ▼a[Sl]▼bNew  York  University▼c2024
■260  1▼aAnn  Arbor▼bProQuest  Dissertations  &  Theses▼c2024
■300    ▼a103  p
■500    ▼aSource:  Dissertations  Abstracts  International,  Volume:  86-04,  Section:  B.
■500    ▼aAdvisor:  Lake,  Brenden  M.
■5021  ▼aThesis  (Ph.D.)--New  York  University,  2024.
■520    ▼aHumans  leverage  compositionality  to  efficiently  learn  new  visual  concepts,  understanding  how  familiar  parts  can  be  combined  to  form  novel  objects.  Moreover,  our  conceptual  knowledge  of  these  novel  objects  is  often  enhanced  by  understanding  how  parts  relate  to  their  functions  and  how  composing  the  parts  relates  to  composing  the  functions.  In  contrast,  popular  computer  vision  models  struggle  to  make  the  same  types  of  inferences,  requiring  more  data  and  generalizing  less  flexibly  than  people  do.In  this  thesis,  we  explore  two  aspects  of  human  visual  learning  and  reasoning.  We  first  study  the  unique  human  capability  to  learn  visual  concepts  across  a  range  of  different  types  of  visual  composition,  examining  how  people  classify  and  generate  a  novel  class  of  abstract  visual  concepts  with  rich  relational  structures.  We  also  examine  how  functions  are  learned  and  represented  by  humans  from  limited  visual  observations  and  the  ability  to  perform  zero-shot  function  compositions  in  various  interaction  conditions.  In  all  studies,  we  compare  humans  with  computational  models  directly  on  the  same  set  of  learning  tasks,  shedding  light  on  whether  computational  models,  including  Bayesian  program  induction  models  and  meta-learning  neural  networks,  can  perform  compositional  visual  concept  and  function  learning  in  the  complete  and  complex  way  humans  do.
■590    ▼aSchool  code:  0146.
■650  4▼aCognitive  psychology
■650  4▼aComputer  science
■650  4▼aPsychology
■653    ▼aBayesian  modeling
■653    ▼aCognitive  modeling
■653    ▼aCompositionality
■653    ▼aConcept  learning
■653    ▼aMeta-learning
■653    ▼aReasoning
■690    ▼a0633
■690    ▼a0984
■690    ▼a0621
■71020▼aNew  York  University▼bCenter  for  Data  Science.
■7730  ▼tDissertations  Abstracts  International▼g86-04B.
■790    ▼a0146
■791    ▼aPh.D.
■792    ▼a2024
■793    ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17163427▼nKERIS▼z이  자료의  원문은  한국교육학술정보원에서  제공합니다.

Preview

Export

ChatGPT Discussion

AI Recommended Related Books


    New Books MORE
    Statistics for the past 3 years. Go to brief

    Подробнее информация.

    • Бронирование
    • не существует
    • моя папка
    • Первый запрос зрения
    • Non-Book Loan Application
    • Nighttime Book Loan Application
    материал
    Reg No. Количество платежных Местоположение статус Ленд информации
    TF12231 전자도서 대출가능 My Folder 부재도서신고 비도서대출신청 야간 도서대출신청

    * Бронирование доступны в заимствований книги. Чтобы сделать предварительный заказ, пожалуйста, нажмите кнопку бронирование

    Books borrowed together with this book

    Related Popular Books

    Available after logging in.