北航数学论坛学术报告

Value of Moment Information in Wasserstein Distributionally Robust Optimization

陈彩华

（南京大学）

报告时间：16：00-17：00，2025-03-21（星期五）

报告地点： 沙河E-802

内容简介：Data-driven optimization often relies on limited sample data, making the trade-off between numerical tractability and information richness a critical consideration. In this paper, we propose a composite ambiguity set that integrates Wasserstein distance with first-order moment information. We begin by quantifying the volume of this ambiguity set under a family of Gaussian distributions. Then, through a non-asymptotic analysis, we evaluate out-of-sample performance, showing that incorporating moment information enhances the robustness of Wasserstein-based distributionally robust optimization. Our findings suggest that with carefully selected parameters, superior performance can be achieved even with limited data. Finally, we develop a tailored algorithm for this framework. Numerical experiments in portfolio selection with real-world data validate the effectiveness of our approach.

报告人简介：陈彩华，国家级青年基金获得者、国家自然科学基金重大项目课题负责人、美国斯坦福大学访问学者，现任南京大学教授、博士生导师、工程管理学院副院长、民建江苏省委大数据与人工智能委员会主任。南京大学理学博士，新加坡国立大学联合培养博士。从事大数据分析与决策、优化算法设计与应用、数据驱动的决策等研究，获华人数学家联盟最佳论文奖(2017、2018)，中国运筹学会青年科技奖(2018)，南京大学青年五四奖章(2019)，南京大学青年五四奖章集体（2024）。

邀请人：韩德仁