Statistics and Data Science Seminar: "Modern Statistical Tools for Network Data: From Resampling to Conformal Prediction"

Speaker: Robert Lunde, University of Michigan

Abstract: Network data, which represent complex relationships between different entities, have become increasingly common in fields ranging from neuroscience to social network analysis. To address key scientific questions in these domains, versatile inferential methods for network-valued data are needed.  In this talk, I will discuss network analogs of two modern statistical methods: the bootstrap and conformal prediction.  In the first part of the talk, I will present results on computation-inference tradeoffs for our fast, randomized linear bootstrap as well as higher-order correctness properties of a more accurate quadratic bootstrap.  In the second part of the talk, I will discuss the properties of conformal prediction for network-assisted regression. While network data are generally dependent, we show that conformal prediction offers similar guarantees to those established in other settings.  This is joint work with Qiaohui Lin, Purnamrita Sarkar, Elizaveta Levina, and Ji Zhu.

Host: Likai Chen and Debashis Mondal