A hybrid singular value thresholding algorithm with diagonal-modify for low-rank matrix recovery

In this paper, a new hybrid singular value thresholding with diagonal-modify algorithm based on the augmented Lagrange multiplier (ALM) method was proposed for low-rank matrix recovery, in which only part singular values were treated by a hybrid threshold operator with diagonalupdate, and which allowed the algorithm to make use of simple arithmetic operation and keep the computational cost of each iteration low. The new algorithm decreased the complexity of the singular value decomposition and shortened the computing time. The convergence of the new algorithm was discussed. Finally, numerical experiments shown that the new algorithm greatly improved the solving efficiency of a matrix recovery problem and saved the calculation cost, and its effect was obviously better than that of the other algorithms mentioned in experiments.