检索结果-内蒙古大学图书馆

arXiv 2024年

作者： Lin, Ruixi You, Yang Department of Computer Science National University of Singapore Singapore

Large language models (LLMs) are good knowledge bases but struggle to perform equally well for all classes in text classification tasks. This paper investigates a fundamental inference-time problem in language models: imbalanced class accuracies. We find what’s underneath the issue is a tendency to over-predict some classes while under-predicting some others. This class accuracy imbalance is difficult to solve from the root via better pre-training or fine-tuning strategies, but we show it can be effectively mitigated via inference-time combinatorial optimization. To this end, we conceptualize and quantify the over- and under-prediction issue as the Contextual Oddity Bias (COBias), and propose the Debiasing as Nonlinear integer programming (DNIP) model to correct in-context learned class probabilities based on minimizing COBias and maximizing overall accuracy, without LLM parameter update. Considering that the DNIP model implicitly contains non-differentiable elements, we therefore use the simulated annealing algorithm to solve it. Extensive evaluations on three LLMs across seven NLP classification tasks in different prompting settings show that DNIP simultaneously achieves significant COBias reduction (-27%) and accuracy improvement (+12%) over the conventional ICL approach, suggesting that inference-time mitigation of class accuracy imbalance is a promising direction to push forward LLM performances. Copyright © 2024, The Authors. All rights reserved.

关键词： integer linear programming

来源：评论

学校读者我要写书评

暂无评论

Deterministic Multistage Constellation Reconfiguration Using integer programming and Sequential Decision-Making Methods

arXiv

引用

arXiv 2024年

作者： Lee, Hang Woon Williams Rogers, David O. Pearl, Brycen D. Chen, Hao Ho, Koki West Virginia University MorgantownWV26506 United States Stevens Institute of Technology HobokenNJ07030 United States Georgia Institute of Technology AtlantaGA30332 United States Department of Mechanical Materials and Aerospace Engineering School of Systems and Enterprises Daniel Guggenheim School of Aerospace Engineering United States

In this paper, we address the problem of reconfiguring Earth observation satellite constellation systems through multiple stages. The Multi-stage Constellation Reconfiguration Problem (MCRP) aims to maximize the total observation rewards obtained by covering a set of targets of interest through the active manipulation of the orbits and relative phasing of constituent satellites. In this paper, we consider deterministic problem settings in which the targets of interest are known a priori. We propose a novel integer linear programming formulation for MCRP, capable of obtaining provably optimal solutions. To overcome computational intractability due to the combinatorial explosion in solving large-scale instances, we introduce two computationally efficient sequential decision-making methods based on the principles of a myopic policy and a rolling horizon procedure. The computational experiments demonstrate that the devised sequential decision-making approaches yield high-quality solutions with improved computational efficiency over the baseline MCRP. Finally, a case study using Hurricane Harvey data showcases the advantages of multi-stage constellation reconfiguration over single-stage and no-reconfiguration scenarios. Copyright © 2024, The Authors. All rights reserved.

关键词： integer linear programming

来源：评论

学校读者我要写书评

暂无评论

Optimization of High-Speed Railway Line Plans for Tourism Products 11th

Optimization of High-Speed Railway Line Plans for Tourism Pr...

引用

11th International Conference on Traffic and Transportation Studies, ICTTS 2024

作者： Lu, Yufei Ke, Yu Yuan, Wuyang School of Transportation Engineering East China Jiaotong University Nanchang330013 China

ISBN: (纸本)9789819796397

With the rapid development of the tourism industry, passenger demand is experiencing significant growth and increasing diversity. The gradual establishment of the "Eight Vertical and Eight Horizontal" high-speed railway network framework presents a prime opportunity for integrating "railway + tourism". This integration requires railway operators to design line plans based on the spatial and temporal distribution characteristics of passenger demand. Furthermore, these plans must be adjusted according to tourism products to provide appropriate transportation services. In response to the passenger demand of high-speed railway, we propose an integer linear programming model for the line plan of multi-day tourism products. We select the high-speed railway in Jiangxi Province as a real-world case to validate the effectiveness of our proposed model. © Beijing Paike Culture Commu. Co., Ltd. 2025.

关键词： integer linear programming

来源：评论

学校读者我要写书评

暂无评论

Scheduling Generative-AI Job DAGs with Model Serving in Data Centers 32

Scheduling Generative-AI Job DAGs with Model Serving in Data...

引用

IEEE/ACM 32nd International Symposium on Quality of Service (IWQoS)

作者： Zheng, Ying Jiao, Lei Xu, Yuedong An, Bo Wang, Xin Li, Zongpeng Fudan Univ Shanghai Peoples R China Univ Oregon Eugene OR USA Nanyang Technol Univ Singapore Singapore Tsinghua Univ Beijing Peoples R China

ISBN: (纸本)9798350350128

Scheduling generative-AI jobs in the edge computing environment faces multiple non-trivial challenges, including the Directed Acyclic Graph (DAG) dependency among tasks, the intrinsic intertwinement between task scheduling and model selection, and the dynamic unpredictable arrival of job DAGs. In this work, we capture all such challenges and formulate a non-linear integer program to optimize the long-term profit of the generative-AI service provider, i.e., service revenue of the admitted jobs minus system costs of executing the tasks contained in such job DAGs. This problem is NP-hard even in the offline setting. To solve it, we first reformulate it into an equivalent schedule selection problem using generated schedules to tackle complex constraints. Then, we design a new online scheduling method through the online primal-dual technique. Experimental results confirm that our approach can increase the total service profit by up to 41.2% compared to existing algorithms.

关键词： integer linear programming

来源：评论

学校读者我要写书评

暂无评论

Multi-Agent Path Planning for Finite Horizon Tasks with Counting Time Temporal Logics 20

Multi-Agent Path Planning for Finite Horizon Tasks with Coun...

引用

IEEE 20th International Conference on Automation Science and Engineering (CASE)

作者： Lv, Peng Li, Shaoyuan Yin, Xiang Shanghai Jiao Tong Univ Dept Automat Shanghai 200240 Peoples R China Shanghai Jiao Tong Univ Key Lab Syst Control & Informat Proc Shanghai 200240 Peoples R China

ISBN: (纸本)9798350358513;9798350358520

In this paper, we consider the multi-agent path planning problem for high-level tasks with finite horizons. In many situations, there is the need to count how many times a sub-task is satisfied in order to achieve the overall task. However, existing temporal logic languages, such as linear temporal logic, may not be efficient in describing such requirements. To address this issue, we propose a new temporal logic language called Counting Time Temporal Logic (CTTL) that extends linear temporal logic by explicitly counting the number of times that some tasks are satisfied. To solve the CTTL path planning problem, we use integer linear programming to encode the satisfaction of the task. We demonstrate that our approach is both sound and complete. To validate our results, we present numerical experiments to show the scalability of the proposed approach. Furthermore, we provide a simulation case study of a team of autonomous robots to illustrate the synthesis procedure.

关键词： integer linear programming

来源：评论

学校读者我要写书评

暂无评论

Complexity of integer programming in Reverse Convex Sets via Boundary Hyperplane Cover

arXiv

引用

arXiv 2024年

作者： Hildebrand, Robert Göß, Adrian Grado Department of Industrial and Systems Engineering Virginia Tech United States Germany

We study the complexity of identifying the integer feasibility of reverse convex sets. We present various settings where the complexity can be either NP-Hard or efficiently solvable when the dimension is fixed. Of particular interest is the case of bounded reverse convex constraints with a polyhedral domain. We introduce a structure, Boundary Hyperplane Cover, that permits this problem to be solved in polynomial time in fixed dimension provided the number of nonlinear reverse convex sets is fixed. Copyright © 2024, The Authors. All rights reserved.

关键词： integer linear programming

来源：评论

学校读者我要写书评

暂无评论

Course Scheduling Made Easier: A User-Friendly Web-Based Timetabling Tool Using PyGLPK 21st

Course Scheduling Made Easier: A User-Friendly Web-Based Tim...

引用

21st International Conference on Cooperative Design, Visualization, and Engineering (CDVE)

作者： Anthony, Barbara M. Mueller, Mark Southwestern Univ Dept Math & Comp Sci 1001 E Univ Ave Georgetown TX 78626 USA

ISBN: (纸本)9783031713149;9783031713156

Course scheduling is a necessary but often challenging task, requiring consideration of university requirements, practical limitations, departmental needs, and faculty preferences. Furthermore, it can be difficult to determine if a proposed schedule satisfies all of the requirements, especially as changes may be made within the process. Clearly articulated schedule requirements can be translated into an integer linear program by operations researchers, and those with domain-specific knowledge can then use linear programming solvers like PyGLPK. However, few faculty members in a university are likely to have that experience, so we developed a web-based tool, including the underlying Python code, that allows for entry of information about courses and the faculty teaching them and creates a schedule consistent with the requirements. More than simply a scheduler, a user can propose changes such as swapping two courses or placing a course at a specific time;the website will either do so, or report that such a change is infeasible. In addition, a user can share the website and file with colleagues, allowing them to better appreciate what is and is not possible for the overall schedule. Recognizing that the precise set of scheduling requirements is often unique to a given university, the source code is made available in a public GitHub repository, providing opportunities for other universities to create a version customized to their needs and structure.

关键词： integer linear programming University timetabling User interface

来源：评论

学校读者我要写书评

暂无评论

Graph-Based Local Planning with Spatiotemporal Risk Assessment For Risk-Bounded and Prediction-Aware Autonomous Driving* 18

Graph-Based Local Planning with Spatiotemporal Risk Assessme...

引用

18th International Conference on Control Automation Robotics and Vision

作者： Ahmad, Abdulrahman Khonji, Majid Al-Sumaiti, Ameena Dias, Jorge Elbassioni, Khaled Khalifa Univ Dept Comp Sci Abu Dhabi U Arab Emirates Khalifa Univ Ctr Autonomous Robot Syst KUCARS Abu Dhabi U Arab Emirates Khalifa Univ Dept Comp & Informat Engn Abu Dhabi U Arab Emirates Khalifa Univ Dept Elect Engn Adv Power Energy Ctr APEC Smart OR Lab Abu Dhabi U Arab Emirates

ISBN: (纸本)9798331518509;9798331518493

Risk-bounded motion planning for autonomous driving in dynamic environments presents significant research challenges. Ensuring continuous navigation towards a destination while making real-time decisions is a nonconvex problem. This paper presents a graph-based local planning method constrained by user-specific driving preference, represented as a risk-bound criterion for motion planning. First, we propose a lattice graph construction method that adheres to the vehicle's curvature constraints. Then, we formulate the trajectory planning problem as an integer-linear programming task, addressed by our novel risk-bounded and prediction-aware constrained shortest path. Our solution accounts for both static and dynamic obstacles in urban settings, adhering to traffic regulations. At the core of our approach is a conservative spatiotemporal risk assessment mechanism, which evaluates collisions considering the uncertain delay from speed control of the ego vehicle and predicted trajectories of dynamic obstacles. We implemented our solution using the CARLA simulator and the ROS2 platform, within a comprehensive framework encompassing global planning, local planning, and vehicle control. The effectiveness of our approach is demonstrated through notable collision avoidance, improved path-tracking, and enhanced risk-bounded planning capabilities.

关键词： integer linear programming

来源：评论

学校读者我要写书评

暂无评论

MILP and Metaheuristic Approaches for HHCRSP with Optional Starting Point: Enhancing Efficiency in Home Healthcare 10

MILP and Metaheuristic Approaches for HHCRSP with Optional S...

引用

10th International Conference on Control, Decision and Information Technologies (CoDIT)

作者： Schwartz, Leo Grunder, Olivier El Hassani, Amir Hajjam Univ Technol Belfort Montbeliard SINERGIES F-90000 Belfort France

ISBN: (纸本)9798350373981;9798350373974

This study presents an innovative extension to Home Healthcare Scheduling and Routing Problem (HHCRSP), introducing an Optional Starting Point (OSP) that allows for personalization of routes from the caregivers' homes directly to patients. We developed a model using Mixed integer linear programming (MILP) and a metaheuristic method inspired by the Greedy Randomized Adaptive Search Procedure (GRASP). This combination aims to make scheduling and routing more efficient, saving time and improving the quality of care for patients at home. Our results show that this new method can significantly enhance how home healthcare is delivered, making it more flexible and effective for both caregivers and patients.

关键词： integer linear programming

来源：评论

学校读者我要写书评

暂无评论

Maximization of Profit and Throughput of EV-Charging Station Integrated with PV and ESS 99

Maximization of Profit and Throughput of EV-Charging Station...

引用

IEEE 99th Vehicular Technology Conference (VTC-Spring)

作者： Kota, Ram Mohan Chowdary Mayank, Jaishree Nayak, Sanjeet Kumar Indian Inst Informat Technol Design & Mfg Dept Comp Sci & Engn Chennai Tamil Nadu India

ISBN: (纸本)9798350387414

Electric vehicles (EVs) provide a promising solution to global pollution by replacing conventional vehicles. Enough public charging stations must be built to encourage the broad adoption of electric vehicles (EVs). Additionally, efficient scheduling and planning of EV arrivals at charging stations with a variety of power sources is essential for improving overall operating efficiency. This paper presents a framework for optimizing both the profit and throughput of a charging station that incorporates a solar photovoltaic system and an energy storage system. We present an integer linear programming model to find the optimal solution. We propose a heuristic strategy to efficiently find near-optimal solutions. Our experimental results demonstrate that the proposed approach can provide a reasonably good solution and for small inputs, the performance deviation is less than 5%. For larger inputs, a comparison of the heuristic and baseline is presented. In certain scenarios, the heuristic performs well, while the baseline exhibits better performance in others. We can use both approaches and select the one that yields better results.

关键词： Electric Vehicle Charging Station Charging Port Scheduling Solar Photovoltaic Energy Storage System integer linear programming

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：