Sun Dec 8th through Sat the 14th, 2019 at Vancouver Convention Center
After reading the author's rebuttal, all four reviewers provide accepting scores. Reviewer 4 increased his/her score despite not being fully convinced by the justification of the low rank assumption on the transition probability matrix. Overall there is agreement among the reviewers that the paper makes an interesting contribution and they value that the authors could run their method in a complex environment like Atari. The AC encourages the authors to include the suggestions made by the reviewers (in particular those from reviewer 3 and 4) for the camera ready version of the paper.