NeurIPS 2019
Sun Dec 8th through Sat the 14th, 2019 at Vancouver Convention Center
Paper ID:1346
Title:Rethinking the CSC Model for Natural Images

This paper presents a unifying framework for understanding both patch-based sparse dictionary learning and covolutional sparse coding. Both are well-studied methods, but there was previously limited understanding of the relationship between them. This paper presents a new perspective on old methods that leads to new insights and helps the authors make adjustments to the convolutional framework to improve its performance.