NeurIPS 2020

ICE-BeeM: Identifiable Conditional Energy-Based Deep Models Based on Nonlinear ICA

Meta Review

All reviewers find this work interesting and solid. I agree that this paper studies an important theoretical question of idenfiability of energy-based models, and thus may be of interest to the community. The experiments are not that strong as reviewers point out, but that is understandable for a theoretical paper. I recommend acceptance.