Sublinear Time Low-Rank Approximation of Positive Semidefinite Matrices

We show how to compute a relative-error low-rank approximation to any positive semidefinite (PSD) matrix in sublinear time, i.e., for any PSD matrix , in time we output a rank- matrix , in factored form, for which , where is the best rank- approximation to . When and are not too large compared to the sparsity of , our algorithm does not need to read all entries of the matrix. Hence, we significantly improve upon previous time algorithms based on oblivious subspace embeddings, and bypass an time lower bound for general matrices (where denotes the number of non-zero entries in the matrix). We prove time lower bounds for low-rank approximation of PSD matrices, showing that our algorithm is close to optimal. Finally, we extend our techniques to give sublinear time algorithms for low-rank approximation of in the (often stronger) spectral norm metric and for ridge regression on PSD matrices.
View on arXiv