Community detection in weighted networks via the profile-pseudo likelihood method

发布时间：2025-03-17 作者：浏览次数：

Speaker:	刘秉辉	DateTime:	2025年3月22日（周六）上午10：00 - 11：00
Brief Introduction to Speaker: 刘秉辉，东北师范大学，教授、博导，统计系主任；主要研究方向为统计机器学习和网络数据分析；在统计学及相关领域发表论文三十余篇，部分成果发表在JASA、AOS、AOAS、AIJ、JMLR、JOE、JBES等上；主持国家自然科学基金多项，入选国家级青年人才计划，担任中国现场统计研究会因果推断分会副理事长、中国现场统计研究会统计交叉科学研究分会副理事长等。		Place:	国交2号楼315会议室
Abstract:In this paper, we consider the issue of community detection in weighted networks. Most methods addressing this issue, particularly those statistical approaches based on likelihood optimization, suffer from a notable drawback: the necessity to specify in advance the particular form of the distribution of edge weights conditional on the community labels. This requirement dictates that algorithms based on likelihood optimization must be custom-tailored exclusively to the specific form of distribution, which exhibits significant limitations in practical applications where the form of distribution is unknown. To address this limitation, we propose two novel methods based on the expectation profile-pseudo likelihood maximization, for community detection in both undirected and directed weighted networks, which are applicable to various types of weighted networks and independent of the specific form of the conditional distribution of the edge weights. In theory, we establish weak and stro...

上一条： Weighted Estimation Method of High Dimension Portfolio Allocation for Time-Varying Stock Market
下一条： Finite type conditions of weakly pseudoconvex hypersurfaces