勷勤数学•专家报告

题      目：Two-way Latent Matching Model for Network Analysis

报  告  人： 王江洲 副研究员  (邀请人：吴琴)

                                          清华大学

时      间： 4月10日  17：00-18:00

地     点：数科院东楼401

报告人简介:

       王江洲，深圳大学数学科学学院，统计与数据科学系 副研究员。主要研究方向：大规模网络数据的统计分析、大规模相依数据的多重检验、机器学习和深度学习等与统计学的交叉研究。目前在统计学领域期刊发表SCI论文十余篇，其中包括：JASA、JMLR、JCGS、JMVA、CSDA、Computational Statistics 和 Stat等国际期刊。主持科研项目：国家自然科学基金青年项目1项、广东省自然科学基金面上项目1项、中国博士后科学基金面上项目和特别资助（站中）项目各1项、参与面上项目1项。入选深圳市“鹏城孔雀计划”特聘岗位C岗。曾多次受邀在ICSA等国际会议上做报告，并担任JMLR, AOAS, Sinica, JCGS, EJS, CSDA和SII等期刊的审稿人。

摘      要：

       We introduce the Two-way Latent Matching Model (TLMM), a new statistical framework for analysing network data that captures the latent matching structures underlying network interactions. The TLMM integrates dimension-reduction techniques from latent factor modelling within an additive formulation that embodies intrinsic symmetry. We establish conditions ensuring model identifiability and develop a maximum likelihood estimation procedure. The resulting estimators are shown to attain the optimal convergence rate and to be asymptotically normal as the network size increases. Extensive numerical experiments confirm the validity of the theoretical results and demonstrate the empirical performance of the proposed algorithms. Applications to the Divvy bike-sharing data and the co-author network further illustrate the model’s practical utility and its ability to uncover interpretable latent structures in real-world networks.

          欢迎老师、同学们参加、交流！