related documents Random matrices: l(1) concentration and dictionary learning with few samples Conference Proceeding