yuejiechi/ScaledGD

Scaled Gradience Descent for Ill-Conditioned Low-Rank Matrix and Tensor Estimation

ScaledGD

The codes implement the experiments described in the paper Accelerating Ill-Conditioned Low-Rank Matrix Estimation via Scaled Gradient Descent, for matrix sensing, matrix completion, robust PCA, Hankel matrix completion respectively.

Each file can run directly in MATLAB, and will generate the figure given in the paper in about 15 minutes.