NeurIPS 2019
Sun Dec 8th through Sat the 14th, 2019 at Vancouver Convention Center
Paper ID:238
Title:Invert to Learn to Invert


		
The idea of iterative inverse models with constant memory is of great interest to solve inverse problems with large number of layers. The proposed method has been successfully applied on a network with 400 layers. Even though the reviewers have raised the question of potential applicability of the technique beyond the domain of MRI reconstruction, there is clearly substantial novel material in the paper that warrants its publication as a poster. Almost all the reviewers have pointed out that the paper is densely written and needs to be revised to add more clarity.