서브메뉴
검색
Learning Dynamic Priority Scheduling Policies with Graph Attention Networks
Learning Dynamic Priority Scheduling Policies with Graph Attention Networks
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20260202105538
- ISBN
- 9798265400574
- DDC
- 658.404
- 저자명
- Wang, Zheyuan.
- 서명/저자
- Learning Dynamic Priority Scheduling Policies with Graph Attention Networks
- 발행사항
- [Sl] : Georgia Institute of Technology, 2022
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2022
- 형태사항
- 167 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 87-05, Section: A.
- 주기사항
- Advisor: Bloch, Matthieu R.;Gombolay, Matthew.
- 학위논문주기
- Thesis (Ph.D.)--Georgia Institute of Technology, 2022.
- 초록/해제
- 요약Resource optimization plays an important role in many real-world scenarios, including health care, manufacturing and services industries, and more. In those resource-constrained environments, effective sequencing and scheduling of workers and jobs has become a necessity for success. Activities must be scheduled to meet various temporal constraints while using the resources available in an efficient manner. Traditional methods for solving scheduling problems are based on dynamic programming and integer programming formulations of the problems, which can be approached with either exact methods that are computationally expensive and hard to scale, or hand-crafted heuristics that can give high-quality solutions but require a combined, herculean effort from computer scientists, operations researchers, and industrial engineers to develop.The aim of this thesis is to develop novel graph attention network-based models to automatically learn scheduling policies for effectively solving resource optimization problems, covering both deterministic and stochastic environments. The policy learning methods utilize both imitation learning, when expert demonstrations are accessible at low cost, and reinforcement learning, when otherwise reward engineering is feasible. By parameterizing the learner with graph attention networks, the framework is computationally efficient and results in scalable resource optimization schedulers that adapt to various problem structures.This thesis addresses the problem of multi-robot task allocation (MRTA) under temporospatial constraints. Initially, robots with deterministic and homogeneous task performance are considered with the development of the RoboGNN scheduler. Then, I develop ScheduleNet, a novel heterogeneous graph attention network model, to efficiently reason about coordinating teams of heterogeneous robots. Next, I address problems under the more challenging stochastic setting in two parts. Part 1) Scheduling with stochastic and dynamic task completion times. The MRTA problem is extended by introducing human co workers with dynamic learning curves and stochastic task execution. HybridNet, a hybrid network structure, has been developed that utilizes a heterogeneous graph-based encoder and a recurrent schedule propagator, to carry out fast schedule generation in multi-round settings. Part 2) Scheduling with stochastic and dynamic task arrival and completion times. With an application in failure-predictive plane maintenance, I develop a heterogeneous graph-based policy optimization (HetGPO) approach to enable learning robust scheduling policies in highly stochastic environments.My research fills the current gap between representation learning and policy learning for solving resource optimization problems by building a unified framework. I further advances the idea of learning to schedule by refining and applying it in more complex and challenging scenarios. Through extensive experiments, the proposed framework has been shown to outperform prior state-of-the-art algorithms in different applications. My research contributes several key innovations regarding designing graph-based learning algorithms in operations research.
- 일반주제명
- Schedules
- 일반주제명
- Aircraft
- 일반주제명
- Scheduling
- 일반주제명
- Integer programming
- 일반주제명
- Teaching methods
- 일반주제명
- Optimization techniques
- 일반주제명
- Graph representations
- 일반주제명
- Robots
- 일반주제명
- Robotics
- 일반주제명
- Industrial engineering
- 일반주제명
- Pedagogy
- 일반주제명
- Education
- 기본자료저록
- Dissertations Abstracts International. 87-05A.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008260126s2022 us c eng d■001000017360507
■00520260202105538
■006m o d
■007cr#unu||||||||
■020 ▼a9798265400574
■035 ▼a(MiAaPQ)AAI32314951
■035 ▼a(MiAaPQ)GeorgiaTech70182
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a658.404
■1001 ▼aWang, Zheyuan.
■24510▼aLearning Dynamic Priority Scheduling Policies with Graph Attention Networks
■260 ▼a[Sl]▼bGeorgia Institute of Technology▼c2022
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2022
■300 ▼a167 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 87-05, Section: A.
■500 ▼aAdvisor: Bloch, Matthieu R.;Gombolay, Matthew.
■5021 ▼aThesis (Ph.D.)--Georgia Institute of Technology, 2022.
■520 ▼aResource optimization plays an important role in many real-world scenarios, including health care, manufacturing and services industries, and more. In those resource-constrained environments, effective sequencing and scheduling of workers and jobs has become a necessity for success. Activities must be scheduled to meet various temporal constraints while using the resources available in an efficient manner. Traditional methods for solving scheduling problems are based on dynamic programming and integer programming formulations of the problems, which can be approached with either exact methods that are computationally expensive and hard to scale, or hand-crafted heuristics that can give high-quality solutions but require a combined, herculean effort from computer scientists, operations researchers, and industrial engineers to develop.The aim of this thesis is to develop novel graph attention network-based models to automatically learn scheduling policies for effectively solving resource optimization problems, covering both deterministic and stochastic environments. The policy learning methods utilize both imitation learning, when expert demonstrations are accessible at low cost, and reinforcement learning, when otherwise reward engineering is feasible. By parameterizing the learner with graph attention networks, the framework is computationally efficient and results in scalable resource optimization schedulers that adapt to various problem structures.This thesis addresses the problem of multi-robot task allocation (MRTA) under temporospatial constraints. Initially, robots with deterministic and homogeneous task performance are considered with the development of the RoboGNN scheduler. Then, I develop ScheduleNet, a novel heterogeneous graph attention network model, to efficiently reason about coordinating teams of heterogeneous robots. Next, I address problems under the more challenging stochastic setting in two parts. Part 1) Scheduling with stochastic and dynamic task completion times. The MRTA problem is extended by introducing human co workers with dynamic learning curves and stochastic task execution. HybridNet, a hybrid network structure, has been developed that utilizes a heterogeneous graph-based encoder and a recurrent schedule propagator, to carry out fast schedule generation in multi-round settings. Part 2) Scheduling with stochastic and dynamic task arrival and completion times. With an application in failure-predictive plane maintenance, I develop a heterogeneous graph-based policy optimization (HetGPO) approach to enable learning robust scheduling policies in highly stochastic environments.My research fills the current gap between representation learning and policy learning for solving resource optimization problems by building a unified framework. I further advances the idea of learning to schedule by refining and applying it in more complex and challenging scenarios. Through extensive experiments, the proposed framework has been shown to outperform prior state-of-the-art algorithms in different applications. My research contributes several key innovations regarding designing graph-based learning algorithms in operations research.
■590 ▼aSchool code: 0078.
■650 4▼aSchedules
■650 4▼aAircraft
■650 4▼aScheduling
■650 4▼aInteger programming
■650 4▼aTeaching methods
■650 4▼aOptimization techniques
■650 4▼aGraph representations
■650 4▼aRobots
■650 4▼aRobotics
■650 4▼aIndustrial engineering
■650 4▼aPedagogy
■650 4▼aEducation
■690 ▼a0771
■690 ▼a0546
■690 ▼a0456
■690 ▼a0515
■71020▼aGeorgia Institute of Technology.
■7730 ▼tDissertations Abstracts International▼g87-05A.
■790 ▼a0078
■791 ▼aPh.D.
■792 ▼a2022
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17360507▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.


