Group Symmetry Enables Faster Optimization in Inverse Problems

Junqi Tang, Guixian Xu

We prove for the first time that, if a linear inverse problem exhibits a group symmetry structure, gradient-based optimizers can be designed to exploit this structure for faster co…