91国产

报告时间：2026年2月27日（周五）下午 14:30-15:30

报告地点：91国产 天赐庄校区精正楼306

报告人：冯瑞，华威大学

报告摘要：

Reciprocity, the tendency of two individuals to form mutual connections, is commonly observed in directed networks. In this talk, we will discuss some results of modeling reciprocity in directed networks with degree heterogeneity, another distinctive feature of real networks, characterizing the variational ability of individuals to make connections. In particular, we first consider the Bernoulli model with reciprocity and its extensions that incorporate node-specific heterogeneity and link-specific reciprocity using covariates. We propose a straightforward inference procedure based on maximum likelihood estimation and discuss the intriguing interplays between non-reciprocal and reciprocal effects in sparse networks. Beyond that, we consider a model that is closely connected with the p1 model, with infinite-dimensional heterogeneity parameters. We propose an estimator for the reciprocal parameters in this model based on a novel conditioning argument. We derive consistency and asymptotic normality for the estimator and address its rate optimality. Simulation studies and two real dataset analyses demonstrate the use of our model.

报告人简介：

冯瑞，华威大学博士生，导师为冷琛雷教授，本科和硕士分别毕业于山东大学和上海交大。主要研究方向为网络数据分析和高维数据分析，已在J. Amer. Statist. Assoc.和Biometrika期刊发表多篇论文。

邀请人：董自康