본문

서브메뉴

A Modern Treatise on Variational Inequality Problems: Structures, Algorithms, and Iteration Complexities- [electronic resource]
A Modern Treatise on Variational Inequality Problems: Structures, Algorithms, and Iteratio...
A Modern Treatise on Variational Inequality Problems: Structures, Algorithms, and Iteration Complexities- [electronic resource]

상세정보

자료유형  
 학위논문파일 국외
최종처리일시  
20240214101218
ISBN  
9798379958060
DDC  
519
저자명  
Huang, Kevin.
서명/저자  
A Modern Treatise on Variational Inequality Problems: Structures, Algorithms, and Iteration Complexities - [electronic resource]
발행사항  
[S.l.]: : University of Minnesota., 2023
발행사항  
Ann Arbor : : ProQuest Dissertations & Theses,, 2023
형태사항  
1 online resource(200 p.)
주기사항  
Source: Dissertations Abstracts International, Volume: 85-02, Section: B.
주기사항  
Advisor: Zhang, Shuzhong.
학위논문주기  
Thesis (Ph.D.)--University of Minnesota, 2023.
사용제한주기  
This item must not be sold to any third party vendors.
초록/해제  
요약In this thesis, we study the variational inequality (VI) problem with the methodology of designing efficient (or optimal) algorithms and analyzing their (sample/gradient) iteration complexities. In particular, we aim to explore the hidden structures in VI that have not been (fully) studied before and use them as insights to guide the development of new optimal algorithms that align with the modern research trends in both VI and optimization. We start from the first-order methods, where acceleration has been established in algorithms such as extra-gradient method, optimistic gradient descent ascent method, and dual extrapolation method. These methods are known as optimal in the sense that they match the lower iteration complexity bounds established for first-order methods in (strongly) monotone VI. We observe that these acceleration schemes in VI, together with the acceleration schemes used in optimization such as Nesterov's acceleration, share a common structure: using additional sequence(s), which we refer to it as ``extra points'', to help improve the convergence of the main sequence. We then propose a general guideline, called the extra-point approach, to construct optimal first-order methods via a more systematic way, which provides flexibility in adopting a variety of extra points/sequences such that the lower bounds take effect. Moving towards high-order methods, research before has relied on using high-order Taylor approximation and an iterative binary-search in solving the subproblems. We show that both of them are not necessary in developing high-order methods, and the key lies in satisfying a high-order Lipschitz bound for any approximation operator used in the subroutine, as well as an appropriate order of regularization to eliminate the needs of binary-search. The proposed unifying framework largely relieves the demand on the complicated analysis derived for different methods and allows us to focus more on the problem structure to design a suitable approximation operator in the algorithm.We also investigate stochastic algorithms for VI, mainly focusing on the stochastic approximation (SA) approach. We propose stochastic extensions of two new first-order methods, which could be viewed as special instances following the aforementioned extra-point approach, and show that optimal iteration complexities can be established for them, in both situations where the stochastic errors are bounded separately or they are reduced together with the deterministic terms. Application is discussed using the example of black-box saddle point problem where even the function values can only be estimated with noises. Using a smoothing technique, we show that by constructing the stochastic zeroth-order gradients, the previous schemes can be readily applied with the guarantee on sample iteration complexities. Another aspect of stochasticity is discussed following the similar line of research, where we study the VI problems with the finite-sum structure. In addition, such finite-sum structure consists of both general vector mappings and gradient mappings. Developments in variance reduced algorithms for both finite-sum optimization and finite-sum VI have been found recently, but none has focused on finite-sum VI with optimization structures. We propose two algorithms for both monotone and strongly monotone VI that explicitly make use of such optimization structure and demonstrate that they indeed serve as a bridge between these two problem classes and are able to perform better than general variance reduced VI algorithms when such structure is actually present. We show that the saddle point reformulation of a finite-sum optimization with finite-sum constraints immediately take the aforementioned forms in VI, where applications are commonly seen in Neyman-Pearson classification in machine learning.Finally, the research in this thesis is extended to non-monotone VI and the solution methods for solving it. Without the monotonicity, another (weaker) global property of the VI problem, the existence of Minty solution, comes to play a central role in the convergence of accelerated projection-type methods. With a slightly worse iteration complexity than the monotone VI, we show that how a general high-order extra-gradient-type method using the concept of the aforementioned approximation can converge. Furthermore, when the existence of Minty solution is no longer assumed, very little can be said in general about the convergence of these projection-type methods. Alternatively, we use these methods as starting points and derive sufficient conditions that characterize various structures of VI where they can converge with guaranteed iteration complexity bounds. This approach allows us to extend our study to potentially broader VI problem classes that have no monotonicity nor Minty solutions.
일반주제명  
Applied mathematics.
일반주제명  
Industrial engineering.
키워드  
Convex optimization
키워드  
Iteration complexities
키워드  
Numerical methods
키워드  
Variational inequalities
기타저자  
University of Minnesota Industrial and Systems Engineering
기본자료저록  
Dissertations Abstracts International. 85-02B.
기본자료저록  
Dissertation Abstract International
전자적 위치 및 접속  
로그인 후 원문을 볼 수 있습니다.

MARC

 008240612s2023      us  |||||||||||||||c||eng  d
■001000016933205
■00520240214101218
■006m          o    d                
■007cr#unu||||||||
■020    ▼a9798379958060
■035    ▼a(MiAaPQ)AAI30526094
■040    ▼aMiAaPQ▼cMiAaPQ
■0820  ▼a519
■1001  ▼aHuang,  Kevin.
■24512▼aA  Modern  Treatise  on  Variational  Inequality  Problems:  Structures,  Algorithms,  and  Iteration  Complexities▼h[electronic  resource]
■260    ▼a[S.l.]:▼bUniversity  of  Minnesota.  ▼c2023
■260  1▼aAnn  Arbor  :▼bProQuest  Dissertations  &  Theses,  ▼c2023
■300    ▼a1  online  resource(200  p.)
■500    ▼aSource:  Dissertations  Abstracts  International,  Volume:  85-02,  Section:  B.
■500    ▼aAdvisor:  Zhang,  Shuzhong.
■5021  ▼aThesis  (Ph.D.)--University  of  Minnesota,  2023.
■506    ▼aThis  item  must  not  be  sold  to  any  third  party  vendors.
■520    ▼aIn  this  thesis,  we  study  the  variational  inequality  (VI)  problem  with  the  methodology  of  designing  efficient  (or  optimal)  algorithms  and  analyzing  their  (sample/gradient)  iteration  complexities.  In  particular,  we  aim  to  explore  the  hidden  structures  in  VI  that  have  not  been  (fully)  studied  before  and  use  them  as  insights  to  guide  the  development  of  new  optimal  algorithms  that  align  with  the  modern  research  trends  in  both  VI  and  optimization.  We  start  from  the  first-order  methods,  where  acceleration  has  been  established  in  algorithms  such  as  extra-gradient  method,  optimistic  gradient  descent  ascent  method,  and  dual  extrapolation  method.  These  methods  are  known  as  optimal  in  the  sense  that  they  match  the  lower  iteration  complexity  bounds  established  for  first-order  methods  in  (strongly)  monotone  VI.  We  observe  that  these  acceleration  schemes  in  VI,  together  with  the  acceleration  schemes  used  in  optimization  such  as  Nesterov's  acceleration,  share  a  common  structure:  using  additional  sequence(s),  which  we  refer  to  it  as  ``extra  points'',  to  help  improve  the  convergence  of  the  main  sequence.  We  then  propose  a  general  guideline,  called  the  extra-point  approach,  to  construct  optimal  first-order  methods  via  a  more  systematic  way,  which  provides  flexibility  in  adopting  a  variety  of  extra  points/sequences  such  that  the  lower  bounds  take  effect.  Moving  towards  high-order  methods,  research  before  has  relied  on  using  high-order  Taylor  approximation  and  an  iterative  binary-search  in  solving  the  subproblems.  We  show  that  both  of  them  are  not  necessary  in  developing  high-order  methods,  and  the  key  lies  in  satisfying  a  high-order  Lipschitz  bound  for  any  approximation  operator  used  in  the  subroutine,  as  well  as  an  appropriate  order  of  regularization  to  eliminate  the  needs  of  binary-search.  The  proposed  unifying  framework  largely  relieves  the  demand  on  the  complicated  analysis  derived  for  different  methods  and  allows  us  to  focus  more  on  the  problem  structure  to  design  a  suitable  approximation  operator  in  the  algorithm.We  also  investigate  stochastic  algorithms  for  VI,  mainly  focusing  on  the  stochastic  approximation  (SA)  approach.  We  propose  stochastic  extensions  of  two  new  first-order  methods,  which  could  be  viewed  as  special  instances  following  the  aforementioned  extra-point  approach,  and  show  that  optimal  iteration  complexities  can  be  established  for  them,  in  both  situations  where  the  stochastic  errors  are  bounded  separately  or  they  are  reduced  together  with  the  deterministic  terms.  Application  is  discussed  using  the  example  of  black-box  saddle  point  problem  where  even  the  function  values  can  only  be  estimated  with  noises.  Using  a  smoothing  technique,  we  show  that  by  constructing  the  stochastic  zeroth-order  gradients,  the  previous  schemes  can  be  readily  applied  with  the  guarantee  on  sample  iteration  complexities.  Another  aspect  of  stochasticity  is  discussed  following  the  similar  line  of  research,  where  we  study  the  VI  problems  with  the  finite-sum  structure.  In  addition,  such  finite-sum  structure  consists  of  both  general  vector  mappings  and  gradient  mappings.  Developments  in  variance  reduced  algorithms  for  both  finite-sum  optimization  and  finite-sum  VI  have  been  found  recently,  but  none  has  focused  on  finite-sum  VI  with  optimization  structures.  We  propose  two  algorithms  for  both  monotone  and  strongly  monotone  VI  that  explicitly  make  use  of  such  optimization  structure  and  demonstrate  that  they  indeed  serve  as  a  bridge  between  these  two  problem  classes  and  are  able  to  perform  better  than  general  variance  reduced  VI  algorithms  when  such  structure  is  actually  present.  We  show  that  the  saddle  point  reformulation  of  a  finite-sum  optimization  with  finite-sum  constraints  immediately  take  the  aforementioned  forms  in  VI,  where  applications  are  commonly  seen  in  Neyman-Pearson  classification  in  machine  learning.Finally,  the  research  in  this  thesis  is  extended  to  non-monotone  VI  and  the  solution  methods  for  solving  it.  Without  the  monotonicity,  another  (weaker)  global  property  of  the  VI  problem,  the  existence  of  Minty  solution,  comes  to  play  a  central  role  in  the  convergence  of  accelerated  projection-type  methods.  With  a  slightly  worse  iteration  complexity  than  the  monotone  VI,  we  show  that  how  a  general  high-order  extra-gradient-type  method  using  the  concept  of  the  aforementioned  approximation  can  converge.  Furthermore,  when  the  existence  of  Minty  solution  is  no  longer  assumed,  very  little  can  be  said  in  general  about  the  convergence  of  these  projection-type  methods.  Alternatively,  we  use  these  methods  as  starting  points  and  derive  sufficient  conditions  that  characterize  various  structures  of  VI  where  they  can  converge  with  guaranteed  iteration  complexity  bounds.  This  approach  allows  us  to  extend  our  study  to  potentially  broader  VI  problem  classes  that  have  no  monotonicity  nor  Minty  solutions.
■590    ▼aSchool  code:  0130.
■650  4▼aApplied  mathematics.
■650  4▼aIndustrial  engineering.
■653    ▼aConvex  optimization
■653    ▼aIteration  complexities
■653    ▼aNumerical  methods
■653    ▼aVariational  inequalities
■690    ▼a0796
■690    ▼a0364
■690    ▼a0546
■71020▼aUniversity  of  Minnesota▼bIndustrial  and  Systems  Engineering.
■7730  ▼tDissertations  Abstracts  International▼g85-02B.
■773    ▼tDissertation  Abstract  International
■790    ▼a0130
■791    ▼aPh.D.
■792    ▼a2023
■793    ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T16933205▼nKERIS▼z이  자료의  원문은  한국교육학술정보원에서  제공합니다.
■980    ▼a202402▼f2024

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
    TF08475 전자도서 마이폴더 부재도서신고 비도서대출신청

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

    해당 도서를 다른 이용자가 함께 대출한 도서

    관련 인기도서

    로그인 후 이용 가능합니다.