서브메뉴
검색
Coordination Among Strategic Agents With Asymmetric Information
Coordination Among Strategic Agents With Asymmetric Information
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20250211153013
- ISBN
- 9798384045076
- DDC
- 621.3
- 저자명
- Wei, Xupeng.
- 서명/저자
- Coordination Among Strategic Agents With Asymmetric Information
- 발행사항
- [Sl] : University of Michigan, 2024
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2024
- 형태사항
- 155 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 86-03, Section: B.
- 주기사항
- Advisor: Anastasopoulos, Achilleas.
- 학위논문주기
- Thesis (Ph.D.)--University of Michigan, 2024.
- 초록/해제
- 요약Multi-agent decision making is an essential task in various fields across communication networks, energy communities, autonomous driving, and more. Agents in these settings observe partial information and take actions to optimize individual objectives. Complexities arise as the system designer must coordinate agents who operate under varying objectives and information, in order to optimize a collective goal. Such coordination is hindered by the strategic interplay, the necessity to infer the unseen, and the intertwinement between strategy and belief formations, making it a formidable challenge. In this thesis, we consider game-theoretic coordination methodologies, including mechanism design, information design, and correlated equilibrium.To begin with, we consider a static resource allocation problem in energy communities in Chapter 2. We propose a mechanism that employs employs specially designed allocation and tax functions to incentivize agents toward maximal social welfare. Given the communication constraints which may exist in real life, the proposed mechanism is revised to support the distributed communication networks, in which the allocation and tax functions depend only on messages from neighborhood. The mechanisms are equipped with a learning algorithm for Nash equilibrium, with which agents are able to find the equilibrium collaboratively.Starting from Chapter 3, we advance to dynamical decision-making systems. Chapter 3 studies a sequential Bayesian learning setting. With the idea of introducing a third-party coordinator to avoid the herding behavior and facilitate the social learning process, a combination of mechanism and information designs is investigated. As the general recommendation mechanism of the coordinator is history-dependent, to settle down the issue brought by accumulative dimension of history, we identify a class of summary-based mechanism and prove its sufficiency for the interest of coordinator. A special suboptimal mechanism is then proposed and demonstrates its effectiveness on avoiding bad herding consequence and an improvement on social welfare.Chapter 4 explores the coordination in Markov games with asymmetric information. Based on the ideas of correlated equilibrium in static games, we develop the idea of perfect correlated equilibrium with the help of a coordination device responsible for collecting public observations and sending out recommended partial strategies. Then, utilizing the special structure of Markov games, a structured perfect correlated equilibrium is proposed, enabling scalable coordination schemes. It can be shown by the example of public investment games that structured perfect correlated equilibrium, as a newly-developed cooperative equilibrium concept, extends the possible outcome set of the existing structured perfect Bayesian equilibrium as a non-cooperative equilibrium concept.
- 일반주제명
- Electrical engineering
- 일반주제명
- Computer engineering
- 일반주제명
- Communication
- 키워드
- Game theory
- 키워드
- Mechanism design
- 기타저자
- University of Michigan Electrical and Computer Engineering
- 기본자료저록
- Dissertations Abstracts International. 86-03B.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008250123s2024 us c eng d■001000017164523
■00520250211153013
■006m o d
■007cr#unu||||||||
■020 ▼a9798384045076
■035 ▼a(MiAaPQ)AAI31631478
■035 ▼a(MiAaPQ)umichrackham005852
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a621.3
■1001 ▼aWei, Xupeng.
■24510▼aCoordination Among Strategic Agents With Asymmetric Information
■260 ▼a[Sl]▼bUniversity of Michigan▼c2024
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2024
■300 ▼a155 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 86-03, Section: B.
■500 ▼aAdvisor: Anastasopoulos, Achilleas.
■5021 ▼aThesis (Ph.D.)--University of Michigan, 2024.
■520 ▼aMulti-agent decision making is an essential task in various fields across communication networks, energy communities, autonomous driving, and more. Agents in these settings observe partial information and take actions to optimize individual objectives. Complexities arise as the system designer must coordinate agents who operate under varying objectives and information, in order to optimize a collective goal. Such coordination is hindered by the strategic interplay, the necessity to infer the unseen, and the intertwinement between strategy and belief formations, making it a formidable challenge. In this thesis, we consider game-theoretic coordination methodologies, including mechanism design, information design, and correlated equilibrium.To begin with, we consider a static resource allocation problem in energy communities in Chapter 2. We propose a mechanism that employs employs specially designed allocation and tax functions to incentivize agents toward maximal social welfare. Given the communication constraints which may exist in real life, the proposed mechanism is revised to support the distributed communication networks, in which the allocation and tax functions depend only on messages from neighborhood. The mechanisms are equipped with a learning algorithm for Nash equilibrium, with which agents are able to find the equilibrium collaboratively.Starting from Chapter 3, we advance to dynamical decision-making systems. Chapter 3 studies a sequential Bayesian learning setting. With the idea of introducing a third-party coordinator to avoid the herding behavior and facilitate the social learning process, a combination of mechanism and information designs is investigated. As the general recommendation mechanism of the coordinator is history-dependent, to settle down the issue brought by accumulative dimension of history, we identify a class of summary-based mechanism and prove its sufficiency for the interest of coordinator. A special suboptimal mechanism is then proposed and demonstrates its effectiveness on avoiding bad herding consequence and an improvement on social welfare.Chapter 4 explores the coordination in Markov games with asymmetric information. Based on the ideas of correlated equilibrium in static games, we develop the idea of perfect correlated equilibrium with the help of a coordination device responsible for collecting public observations and sending out recommended partial strategies. Then, utilizing the special structure of Markov games, a structured perfect correlated equilibrium is proposed, enabling scalable coordination schemes. It can be shown by the example of public investment games that structured perfect correlated equilibrium, as a newly-developed cooperative equilibrium concept, extends the possible outcome set of the existing structured perfect Bayesian equilibrium as a non-cooperative equilibrium concept.
■590 ▼aSchool code: 0127.
■650 4▼aElectrical engineering
■650 4▼aComputer engineering
■650 4▼aCommunication
■653 ▼aGame theory
■653 ▼aMulti-agent system
■653 ▼aMechanism design
■653 ▼aInformation design
■653 ▼aMarkov decision process
■653 ▼aBayesian learning
■690 ▼a0544
■690 ▼a0501
■690 ▼a0464
■690 ▼a0459
■71020▼aUniversity of Michigan▼bElectrical and Computer Engineering.
■7730 ▼tDissertations Abstracts International▼g86-03B.
■790 ▼a0127
■791 ▼aPh.D.
■792 ▼a2024
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17164523▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.


