NeurIPS 2019
Sun Dec 8th through Sat the 14th, 2019 at Vancouver Convention Center
Paper ID:5204
Title:Practical Two-Step Lookahead Bayesian Optimization


		
The reviewers mainly agreed that this paper made practically useful and theoretically supported contributions. There was disagreement about the novelty of these contributions. I was convinced by the authors' argument that their central contribution, a stochastic gradient estimator which is both efficient and unbiased, is a worthwhile contribution. In addition, the authors responded convincingly regarding both additional results that they would be including, and comparisons to existing methods. They will also be including source code, and additional experiments on batch experiments in the appendix.