Community Detection in Dynamic Stochastic Block Models for Discrete-Time Temporal Networks

发布时间：2023-05-17 作者：77779193永利官网浏览次数：

Speaker:	刘秉辉教授	DateTime:	2023年5月22日(周一)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；国务院学位委员会第八届统计学学科评议组秘书、东师统计学一流学科工作委员会委员/秘书长；主要研究方向为应用统计、机器学习和网络数据分析；在统计学、计算机&人工智能、计量经济学领域期刊发表SCI论文近三十篇，部分成果发表在：统计学顶级期刊Journal of the American Statistical Association、Annals of Statistics，计算机&人工智能顶级期刊Artificial Intelligence、Journal of Machine Learning Research，计量经济学顶级期刊Journal of Econometrics；以及领域重要期刊Journal of Business & Economic Statistics、Annals of Applied Statistics等；获国家天元数学东北中心优秀青年学者奖励计划（第一类第一层次）资助；获评吉林省第八批拔尖创新人才（第三层次）；主持国家自然科学基金青年项目1项和面上项目2项、中央高校基本科研业务费青年拔尖人才项目、吉林省科技厅重点实验室专项项目、吉林省重点教改项目等；参与国家自然科学基金重点项目、科技部重点研发计划项目等。		Place:	腾讯会议（会议号：405-934-094）
Abstract:In recent years, the detection of communities in discrete-time temporal networks under dynamic stochastic block models has garnered significant attention. These models combine static stochastic block models with Markov chains of community variables, which depict the evolution of these variables. However, fitting a dynamic stochastic block model by maximizing its likelihood function is a challenging task, especially for large-scale networks, as community variables are both potential and dynamic. This method offers advantages in computational efficiency and accuracy of community detection, while also providing a solid theoretical guarantee of convergence and consistency. Furthermore, we extend the dPPL method to handle discrete-time dynamic networks with degree heterogeneity. Finally, through extensive simulation results and real data analyses, we demonstrate the practicality and advantages of the proposed methods.

上一条： Dynamical magneto-rotational instability
下一条：是否每个可数无穷图都有一个不友好剖分？