Reviews: On Fast Leverage Score Sampling and Optimal Learning

After further thought, however, I have updated my review to a 5. While this works seems to have very high potential, the current draft has several important shortcomings: (1) It is not yet polished enough for broad consumption. Significant revisions would be necessary to increase clarity. Summary: This paper presents a novel algorithm for estimating leverage scores for kernel matrices. The running time of this algorithm improves upon state of the art.