题目：Recent Developments in Recommendation System and Distributed Statistical Inference

主讲人：上海交通大学数学科学学院 毛晓军副教授

主持人：西南财经大学统计学院 常晋源教授

时间1：2024年3月13日（周三）上午9:30-12:00

地点1：西南财经大学柳林校区通博楼304

时间2：2024年3月14日（周四）上午9:30-12:00 下午14:00-16:30

地点2：西南财经大学光华校区光华楼10楼1003

报告摘要：

Topic 1：Matrix Completion under Several Missing Mechanisms

Matrix completion stands out as a crucial technique within recommendation systems. While current research has successfully furnished robust theoretical assurances, the complexity of the missing mechanism intensifies with the expansion of matrix dimensions. This presentation introduces multiple strategies tailored to address diverse missing mechanisms. The incorporation of covariates in matrix completion not only refines precision but also lends interpretability to the associated covariates. Addressing complex low-rank missing patterns, matrix completion under low-rank missing mechanisms proves to be an effective solution. Moreover, the utilization of a distributed approach in fast median matrix completion enhances the computational speed of matrix completion algorithms. Lastly, matrix completion with model-free weighting enables the management of general nonuniform missing structures by regulating a balancing error.

Topic 2：Distributed Statistical Inference under Several Scenarios

Distributed learning has undergone thorough examination and practical application across diverse machine-learning contexts. This presentation commences by revisiting prominent distributed learning frameworks, such as Divide-and-Conquer and Communication-efficient Surrogate Likelihood (CSL). Addressing challenges related to sparsity, robustness, and missing data, we introduce distributed statistical inference methods tailored for various scenarios. Subsequently, we delve into fundamental decentralized consensus algorithms and engage in a discussion about decentralized distributed statistical learning approaches.

主讲人简介：

毛晓军，研究领域包括分布式统计推断，推荐系统和高维数据分析。主要研究成果发表于JASA, JMLR, IEEE TIT, IEEE TSP, IEEE TIFS, ICML,《管理世界》等。入选中国科协青年人才托举工程，上海市启明星计划，上海市扬帆计划。目前是国际重要学术期刊Journal of Multivariate Analysis的Early Career Advisory Board成员。先后主持国自然面上项目和青年基金，参与国家自然科学基金重大研究计划等。