Objective Operational Learning and Applications

客观操作学习与应用

• 批准号: 1201085
• 负责人: Andrew Lim
• 金额: $ 30万
• 依托单位: University of California-Berkeley
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2012
• 资助国家: 美国
• 起止时间: 2012-03-15 至 2016-02-29
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1201085&HistoricalAwards=false
• 关键词: Objective; Operational; Learning; Applications

Public Abstract: "Objective Operational Learning and Applications"Andrew Lim and J. George ShanthikumarThe research objective of this award is to study a general approach called "Objective Operational Learning" for estimating an objective function and optimizing a stochastic system. The essential feature of this approach is that it is a hybrid between classical model-based approaches to stochastic optimization and purely non-parametric data driven methods that require minimal assumptions. Its advantage is that it allows the decision maker to incorporate structural knowledge of the system (which may only be partially correct) into the initial model, but to become increasingly data-driven and less dependent on these initial assumptions as the data set increases in size. The goal of the research is to show how objective operational learning can be applied to problems of interest in operations research and management science, and to establish theoretical properties such as small sample performance and asymptotic convergence of this approach.If successful, the research will lead to the development of new data driven methods for optimizing stochastic systems that perform well when the size of the data set is small but with attractive large sample properties, and insight into how these methods can be used through a number of case studies involving applied problems of interest to industry. Applications of interest include (but are not restricted to) staff and patient scheduling applications in healthcare and service systems, pricing and revenue management problems, and inventory control. Outcomes of research will be disseminated through journal and conference publications and research presentations, undergraduate research opportunities, and advanced undergraduate and graduate level courses and seminars.

公开摘要：“客观运算学习和应用“Andrew Lim和J.乔治ShanthiquarThe研究目标的这个奖项是研究一个通用的方法称为“客观运算学习”估计一个目标函数和优化一个随机系统。这种方法的基本特征是，它是一个混合的经典模型为基础的方法，随机优化和纯粹的非参数数据驱动的方法，需要最少的假设。它的优点是，它允许决策者将系统的结构知识（可能只是部分正确）纳入初始模型，但随着数据集大小的增加，它会变得越来越受数据驱动，越来越不依赖于这些初始假设。研究的目标是展示客观运算学习如何应用于运筹学和管理科学中感兴趣的问题，并建立这种方法的小样本性能和渐近收敛等理论性质，如果成功这项研究将导致开发新的数据驱动方法，用于优化随机系统，当数据集的大小很小时，具有吸引人的大样本特性，并通过一些涉及工业应用问题的案例研究深入了解如何使用这些方法。感兴趣的应用程序包括（但不限于）医疗保健和服务系统中的员工和患者调度应用程序，定价和收入管理问题以及库存控制。研究成果将通过期刊和会议出版物和研究报告，本科生研究机会以及高级本科生和研究生课程和研讨会进行传播。