NeurIPS 2019
Sun Dec 8th through Sat the 14th, 2019 at Vancouver Convention Center
Paper ID:2257
Title:Thinning for Accelerating the Learning of Point Processes

This paper proposes a method to efficiently learn the parameters of point processes by using thinning to estimate the parameters and gradients. The paper also develops the theory of the bias and variance of the proposed estimators. Finally, the proposed method is validated experimentally. Overall, the reviewers were quite positive about this paper saying it was well written and that the problem is of practical interest to the ML community. Additionally, the actual method and results seem significant and can open up new research areas. The authors seem to have addressed the major concerns of the reviewers in their response. The reviewers have provided the authors with good feedback to make their paper better and the authors should incorporate those suggestions into the camera-ready version.