NeurIPS 2020

CHIP: A Hawkes Process Model for Continuous-time Networks with Scalable and Consistent Estimation


Meta Review

The reviews are positive: The paper addresses an interesting problem, proposes a reasonably novel model, is well-executed, and makes an entertaining read. Although the reviewers point out a number of concerns (e.g. the dyadic independence assumption that will generally be unrealistic), the opinions and scores support publication. To my mind, the main weakness is perhaps that combining network models and Hawkes processes has been fashionable for a few years now, and although the model is technically novel, I have not found anything in the paper that I would consider unexpected.