Paper Accepted at AISTATS 2021

We introduce a novel modeling framework for point processes called IMTPP (Intermittently-observed Marked Temporal Point Processes), which characterizes the dynamics of both observed and missing events as two coupled MTPPs, conditioned on the history of previous events. Our model of generating missing events depends both on the previously occurred missing (or unobserved) events –MNAR kind– as well as the previously occurred observed events. (Preprint coming up soon)