NeurIPS 2019
Sun Dec 8th through Sat the 14th, 2019 at Vancouver Convention Center
Paper ID:1224
Title:Rethinking Generative Mode Coverage: A Pointwise Guaranteed Approach


		
This paper received good scores overall, with a bit of disparity between reviews. There is consensus that it makes novel contributions in the field latent-variable generative models, providing a metric for mode-collapse and offering a potential solution via a mixture of generators. It also provides a theoretical analysis. The weaknesses of this submission are the experiments and related works, which could be improved to substantially increase the strength of the paper.