Because the algorithm of Robust PCA need iteration to converge, the speed processed in CPU is slow. We run the Robust PCA on GPU,whose speed will promote 10 times.
inexact_augmented_lagrange_multiplier.py : include two function to achieve the iteration of RobustPCA include CPU version (inexact_augmented_lagrange_multiplier) and tensorflow version(inexact_augmented_lagrange_multiplier_tf).
if you want to use the GPU version function, please replace the 74th code with A, E = inexact_augmented_lagrange_multiplier_tf([X:,:sz])