Streaming Sparse Principal Component Analysis

This paper considers estimating the leading k principal components with at most s non-zero attributes from p-dimensional samples collected sequentially in memory limited environments. We develop and a