This repository is for our paper:

[1] HanQin Cai, Jialin Liu, and Wotao Yin. Learned Robust PCA: A Scalable Deep Unfolding Approach for High-Dimensional Outlier Detection. In Advances in Neural Information Processing Systems, 34: 16977-16989, 2021.

Given $Y = X + S$ where $X$ is the underlying low-rank matrix and $S$ is the sparse ourlier matrix, robust principal component analysis (RPCA) aims to recover $X$ and/or $S$ from the observed $Y$, depending on the application.

• synthetic_data_exp involves our codes for the synthetic-data experiments¹.

• Enter synthetic_data_exp and run testing_codes_matlab.m directly.
• The test script will call a trained model stored in synthetic_data_exp/trained_models.

• Enter synthetic_data_exp and run training_codes.py directly.
• The training script will write the model into a .mat file that the test script can load.

• Testing codes: MATLAB (>= 2017b)
• Training codes: CUDA 11.0; pytorch 1.7.1