본문

서브메뉴

Implicit Bias of Deep Learning Optimization: A Mathematical Examination
Implicit Bias of Deep Learning Optimization: A Mathematical Examination
Implicit Bias of Deep Learning Optimization: A Mathematical Examination

상세정보

자료유형  
 학위논문 서양
최종처리일시  
20250211152757
ISBN  
9798384466994
DDC  
004
저자명  
Lyu, Kaifeng.
서명/저자  
Implicit Bias of Deep Learning Optimization: A Mathematical Examination
발행사항  
[Sl] : Princeton University, 2024
발행사항  
Ann Arbor : ProQuest Dissertations & Theses, 2024
형태사항  
392 p
주기사항  
Source: Dissertations Abstracts International, Volume: 86-04, Section: B.
주기사항  
Advisor: Arora, Sanjeev.
학위논문주기  
Thesis (Ph.D.)--Princeton University, 2024.
초록/해제  
요약Deep learning has achieved remarkable success in recent years, yet training neural networks often involves a delicate combination of guesswork and hyperparameter tuning. A critical aspect of this process is the "implicit bias" of optimization methods, where minor changes in the optimization setup-without affecting the small training loss at convergence-can drastically shift the solution to which the model converges, thereby affecting test performance. This dissertation presents a collection of results that mathematically characterize this implicit bias in various training regimes.The first part of this dissertation explores how gradient descent, even without explicit regularization, can converge to solutions that maximize the margin. Previous results have established the first-order optimality of margin for homogeneous neural networks in general, but the global optimality of margin is not guaranteed due to their non-convex nature. This dissertation provides in-depth theoretical analyses when data has simple structures: for linearly separable data, we present both positive and negative results on whether the global optimality of margin can be attained. Furthermore, we show how this margin-based view can be used to explain interesting generalization phenomena in training neural networks with or without explicit regularization, including the simplicity bias and grokking phenomena.The second part of the dissertation presents two results that capture the implicit biases induced by finite learning rate. Many existing analyses, including the margin-based ones in the first part, describe implicit biases that hold even when the learning rate is infinitesimal. However, practical implementations use finite learning rates, which have been empirically observed to benefit generalization. We analyze how full-batch GD with finite learning rates, combined with key training components like normalization layers and weight decay, create a bias towards flatter minima, which are positively correlated with better generalization. Additionally, we study the implicit bias in stochastic optimization and derive rigorous approximations for the dynamics of adaptive gradient methods like Adam and RMSprop via Stochastic Differential Equations (SDEs) to capture the effect of finite learning rates. Based on this, we also derive the square root scaling rule as a practical guideline for adjusting the optimization hyperparameters of adaptive gradient methods when changing batch size.
일반주제명  
Computer science
일반주제명  
Applied mathematics
키워드  
Deep learning
키워드  
Hyperparameter tuning
키워드  
Finite learning rate
기타저자  
Princeton University Computer Science
기본자료저록  
Dissertations Abstracts International. 86-04B.
전자적 위치 및 접속  
로그인 후 원문을 볼 수 있습니다.

MARC

 008250123s2024        us                              c    eng  d
■001000017163821
■00520250211152757
■006m          o    d                
■007cr#unu||||||||
■020    ▼a9798384466994
■035    ▼a(MiAaPQ)AAI31555969
■040    ▼aMiAaPQ▼cMiAaPQ
■0820  ▼a004
■1001  ▼aLyu,  Kaifeng.
■24510▼aImplicit  Bias  of  Deep  Learning  Optimization:  A  Mathematical  Examination
■260    ▼a[Sl]▼bPrinceton  University▼c2024
■260  1▼aAnn  Arbor▼bProQuest  Dissertations  &  Theses▼c2024
■300    ▼a392  p
■500    ▼aSource:  Dissertations  Abstracts  International,  Volume:  86-04,  Section:  B.
■500    ▼aAdvisor:  Arora,  Sanjeev.
■5021  ▼aThesis  (Ph.D.)--Princeton  University,  2024.
■520    ▼aDeep  learning  has  achieved  remarkable  success  in  recent  years,  yet  training  neural  networks  often  involves  a  delicate  combination  of  guesswork  and  hyperparameter  tuning.  A  critical  aspect  of  this  process  is  the  "implicit  bias"  of  optimization  methods,  where  minor  changes  in  the  optimization  setup-without  affecting  the  small  training  loss  at  convergence-can  drastically  shift  the  solution  to  which  the  model  converges,  thereby  affecting  test  performance.  This  dissertation  presents  a  collection  of  results  that  mathematically  characterize  this  implicit  bias  in  various  training  regimes.The  first  part  of  this  dissertation  explores  how  gradient  descent,  even  without  explicit  regularization,  can  converge  to  solutions  that  maximize  the  margin.  Previous  results  have  established  the  first-order  optimality  of  margin  for  homogeneous  neural  networks  in  general,  but  the  global  optimality  of  margin  is  not  guaranteed  due  to  their  non-convex  nature.  This  dissertation  provides  in-depth  theoretical  analyses  when  data  has  simple  structures:  for  linearly  separable  data,  we  present  both  positive  and  negative  results  on  whether  the  global  optimality  of  margin  can  be  attained.  Furthermore,  we  show  how  this  margin-based  view  can  be  used  to  explain  interesting  generalization  phenomena  in  training  neural  networks  with  or  without  explicit  regularization,  including  the  simplicity  bias  and  grokking  phenomena.The  second  part  of  the  dissertation  presents  two  results  that  capture  the  implicit  biases  induced  by  finite  learning  rate.  Many  existing  analyses,  including  the  margin-based  ones  in  the  first  part,  describe  implicit  biases  that  hold  even  when  the  learning  rate  is  infinitesimal.  However,  practical  implementations  use  finite  learning  rates,  which  have  been  empirically  observed  to  benefit  generalization.  We  analyze  how  full-batch  GD  with  finite  learning  rates,  combined  with  key  training  components  like  normalization  layers  and  weight  decay,  create  a  bias  towards  flatter  minima,  which  are  positively  correlated  with  better  generalization.  Additionally,  we  study  the  implicit  bias  in  stochastic  optimization  and  derive  rigorous  approximations  for  the  dynamics  of  adaptive  gradient  methods  like  Adam  and  RMSprop  via  Stochastic  Differential  Equations  (SDEs)  to  capture  the  effect  of  finite  learning  rates.  Based  on  this,  we  also  derive  the  square  root  scaling  rule  as  a  practical  guideline  for  adjusting  the  optimization  hyperparameters  of  adaptive  gradient  methods  when  changing  batch  size.
■590    ▼aSchool  code:  0181.
■650  4▼aComputer  science
■650  4▼aApplied  mathematics
■653    ▼aDeep  learning
■653    ▼aHyperparameter  tuning
■653    ▼aFinite  learning  rate
■690    ▼a0984
■690    ▼a0800
■690    ▼a0364
■71020▼aPrinceton  University▼bComputer  Science.
■7730  ▼tDissertations  Abstracts  International▼g86-04B.
■790    ▼a0181
■791    ▼aPh.D.
■792    ▼a2024
■793    ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17163821▼nKERIS▼z이  자료의  원문은  한국교육학술정보원에서  제공합니다.

미리보기

내보내기

chatGPT토론

Ai 추천 관련 도서


    신착도서 더보기
    최근 3년간 통계입니다.

    소장정보

    • 예약
    • 소재불명신고
    • 나의폴더
    • 우선정리요청
    • 비도서대출신청
    • 야간 도서대출신청
    소장자료
    등록번호 청구기호 소장처 대출가능여부 대출정보
    TF11087 전자도서 대출가능 마이폴더 부재도서신고 비도서대출신청 야간 도서대출신청

    * 대출중인 자료에 한하여 예약이 가능합니다. 예약을 원하시면 예약버튼을 클릭하십시오.

    해당 도서를 다른 이용자가 함께 대출한 도서

    관련 인기도서

    로그인 후 이용 가능합니다.