NeurIPS 2020

Fast Matrix Square Roots with Applications to Gaussian Processes and Bayesian Optimization

Meta Review

After multiple discussions with regards to the experimental evaluation and the use of a hard coded limit J=200 for MINRES, all the reviewers reached a consensus on accepting this paper as they believe it opens up a different direction for scalable GPs and related areas and it provides interesting theoretical analyses. I agree with the reviewers and recommend acceptance. I strongly recommend the authors discuss this thoroughly in their final version along with the additional material presented to the AC via CMT.