Frontiers of Statistical Network Analysis: Inference, Tensors and Beyond

Overview

Network data present unique structures and distinct analytical challenges. Earlier efforts in statistical network analysis focused on modeling and parameter estimation. The past decade, known as a “golden age” for this field, has witnessed a surge in innovative methods and theoretical developments. The recent boom in data science also gives rise to numerous categories of complex networks where interactions among a set of entities are polyadic or non-linear, which are collectively referred to as tensors, or higher-order networks. Moreover, we witness the increasingly interdisciplinary collaborations between network analysis and applied fields.

The workshop will feature three main parts:

• Network Inference: The last decade has seen tremendous strides in both parametric and nonparametric network inference. Topics will include model selection, asymptotic distribution of estimators, and scalable nonparametric methods.
• Higher-Order Networks and Tensors: Tensors, multi-way generalizations of matrices, are natural tools for encoding these networks. This section will explore tensor decomposition techniques, statistical modeling, and the computational challenges inherent in tensor methods.
• Interdisciplinary Applications: This part will highlight how network techniques can engage with other areas such as causal inference, epidemiology, and differential privacy, fostering a broader understanding and inspiring new research directions.

The goal of this workshop is three-fold: to advance the field by exchanging ideas and deepening understanding; to facilitate future research by discussing challenges and establishing collaboration opportunities; and to benefit the local Singaporean audience and other participants through lectures from leading experts and emerging star scholars.