Bilinear Adaptive Vector Approximate Message Passing (BAdVAMP) IEEE, arxive

BAdVAMP is an algorithm based on VAMP to recover b and x from measurements y=A(b)x + w, where w is white Gaussian noise and A(b) is of the form A(b) = A0 + b1*A1 + … + bk * Ak with known matrices A0, …, Ak.

There are two example files provided,

demo_CSMU.m : Compressive Sensing with Matrix Uncertainty (CSMU) problem

demo_DL.m : Dictionary Learning problem

Note: before runnning the files make sure that the gampmatlab toolbox is added to Matlab’s path.