讲坛地点：舜耕校区鱼乐达人 315会议室

报告时间：2026年10月16日（星期五）16：00—18：00

主办单位：鱼乐达人网页版

协办单位：科研处，统计交叉科学研究中心，中加项目管理委员会

主持人：于渊，鱼乐达人

1.报告一

题目：Ball Impurity: Measuring Heterogeneity in General Metric Spaces

报告人：潘文亮

摘要：Data in various domains, such as neuroimaging and network data analysis, often come in complex forms without possessing a Hilbert structure. The complexity necessitates innovative approaches for effective analysis. We propose a novel measure of heterogeneity, ball impurity, which is designed to work with complex non-Euclidean objects. Our approach extends the notion of impurity to general metric spaces, providing a versatile tool for feature selection and tree models. The ball impurity measure exhibits desirable properties, such as the triangular inequality, and is computationally tractable, enhancing its practicality and usefulness. Extensive experiments on synthetic data and real data from the UK Biobank validate the efficacy of our approach in capturing data heterogeneity. Remarkably, our results compare favorably with state-of-the-art methods in metric spaces, highlighting the potential of ball impurity as a valuable tool for addressing complex data analysis tasks.

特邀专家简介：潘文亮，现任中国科学院数学与系统科学研究院副研究员及博士生导师，专注于统计学习算法、医学图像数据分析和度量空间的非参数方法等领域研究。在Annals of Statistics、Journal of the American Statistical Association、TPAMI等统计学及人工智能顶级杂志上发表了30篇以上学术论文，获得2022年教育部高等学校科学研究优秀成果自然科学类二等奖（排名第二）。主持的科研项目涵盖国家自然科学基金委青年基金B类、面上项目等。同时，担任北京生物医学统计与数据管理研究会副理事长，以及中国现场统计研究会统计交叉科学研究分会副秘书长。

2.报告二

题目：Causal Inference of Sequential Treatment Effects under Flexible Designs

报告人：黄辉

摘要：Estimating causal effects under general treatment-sequence designs is challenging in panel studies, especially when treatment assignments are endogenous and treatment sequences vary substantially across units. To address these challenges, we introduce Design-adaptive Average Treatment effects on the Treated (DATTs), which characterize causal effects associated with the observed treatment sequences without imposing strong restrictions on treatment designs. To estimate DATTs, we introduce an overlapping factor model that jointly models outcomes across different treatment sequences, leveraging their shared structure to borrow information across treated subpopulations for recovering counterfactual outcomes. The proposed factor model exists under mild conditions, and its formulation accommodates heterogeneity induced by endogeneity, captures long-run treatment-outcome dependence, and adapts its complexity to the realized treatment design. We develop an alternating minimization algorithm for model estimation and establish statistical guarantees for the resulting DATT estimators, allowing for general treatment designs without requiring the panel length to diverge. Through simulations, we demonstrate the superior performance of the proposed method over existing competing approaches. Finally, we apply our method to analyze a panel study of COVID-19 vaccination with a complex non-staggered treatment design, illustrating its practical utility in a real-world application.

特邀专家简介：黄辉博士，2004年于中国科学技术大学获得统计学学士学位，2010年于马里兰大学获得统计学博士学位，目前为中国人民大学统计学院教授、博士生导师。2015年入选国家级青年人才计划；研究方向主要集中于空间统计学、时空数据分析、空间流行病等领域；在国际顶级统计学、交叉学科期刊发表论文多篇，先后主持并完成基金委青年项目、面上项目和科技部国家重点研发计划子课题，主持基金委重点项目子课题，并参与其他重大项目多项。

3.报告三

题目：Improving time series estimation and prediction via transfer learning

报告人：朱倩倩

摘要：There are many time series in the literature with high dimension yet limited sample sizes, such as macroeconomic variables, and it is almost impossible to obtain efficient estimation and accurate prediction by using the corresponding datasets themselves. This paper fills the gap by introducing a novel representation-based transfer learning framework for vector autoregressive models, and information from related source datasets with rich observations can be leveraged to enhance estimation efficiency through representation learning. A two-stage regularized estimation procedure is proposed with well established non-asymptotic properties, and algorithms with alternating updates are suggested to search for the estimates. Our transfer learning framework can handle time series with varying sample sizes and asynchronous starting and/or ending time points, thereby offering remarkable flexibility in integrating information from diverse datasets.  Simulation experiments are conducted to evaluate the finite-sample performance of the proposed methodology, and its usefulness is demonstrated by an empirical analysis on 20 macroeconomic variables from Japan and another nine countries.

特邀专家简介：朱倩倩，上海财经大学统计与数据科学学院教授，博士生导师，入选国家级青年人才项目、上海市“浦江人才”团队项目和“晨光计划”项目，博士毕业于香港大学。研究方向为时间序列分析、金融计量、时空数据建模，研究方法主要应用于宏观经济变量预测和金融风险度量评估。研究成果主要发表在Journal of the Royal Statistical Society, Series B、Journal of Econometrics、Journal of Business & Economic Statistics、Econometric Theory及Statistica Sinica等国际权威期刊上。主持国家自然科学基金面上项目和青年项目，合著《数理统计》教材，获评上海财经大学“我心目中的好老师”和博士研究生“学术之星”指导老师。