报告题目: Community Detection in Complex Networks
主讲人:刘秉辉
报告摘要:The detection of communities in complex networks is a fundamental aspect of complex network studies that has garnered significant attention in recent years. The stochastic block model and its variants have emerged as the most popular statistical models for characterizing community structure,serving as the foundation for many complex network constructions. However, ftting a stochastic block model for complex network by maximizing its likelihood function is often a challenging task,especially for large-scale networks. To address this issue, we have developed a strategy called "decoupling rows and columns" to effectively solve the fitting problems associated with certain complex network models. This strategy involves the creation and optimization of a well-designed profile-pseudo likelihood, which serves as a tractable substitute for the likelihood function. By utilizing this strategy, we are able to achieve computational efficiency and accurate community detection, while also providing a solid theoretical guarantee of convergence and consistency. Through extensive simulations and analyses of real-world data, we demonstrate the practicality and advantages of our proposed strategy in diverse complex networks.
刘秉辉简介：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、计算机&人工智能、计量经济学期刊发表论文三十余篇，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、国家天元数学东北中心优秀青年学者、吉林省拔尖创新人才；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；与中国联通公司、长春市长公开电话办公室等单位合作，主持大数据产品开发、大数据培训若干。
报告时间:2023年12月5日上午8:30
报告地点:综合楼第二阶梯教室
主办单位:数学与统计学院/大数据科学研究院/科研处
