CSE DSI Machine Learning Seminar with Tianxi Li (Statistics, U of MN)

In this talk, Li will showcase a series of recent advancements in network analysis developed by our team using cutting-edge nonparametric inference techniques. The talk begins with an overview of fundamental network data concepts and an essential data processing approach for distilling valuable insights from raw network datasets. He then introduces an aggregation learning strategy, designed to adaptively achieve optimal model performance. The last part of the talk focuses on a motif-sampling inference approach, laying the foundation for broad-scope inference tasks across multiple networks and network populations. All methods presented are grounded in rigorous statistical principles and optimized for computational efficiency. He will illustrate the practical impact of these tools through their applications in analyzing social patterns in collaboration networks, predicting missing links, and unraveling complex biological mechanisms. This talk aims to highlight the significant strides and wide-ranging applications of statistical network analysis, marking its vital role in recent statistical learning research.