Sparse PCA via Covariance Thresholding

In sparse principal component analysis we are given noisy observations of a lowrank matrix of dimension n p and seek to reconstruct it under additional sparsity assumptions.