%0 Generic
%A Wilson, JT
%A Hutter, F
%A Deisenroth, MP
%C Montreal, QC, Canada
%D 2018
%E Bengio, S
%E Wallach, H
%E Larochelle, H
%E Grauman, K
%E CesaBianchi, N
%E Garnett, R
%F discovery:10083559
%I Neural Information Processing Systems (NIPS)
%T Maximizing acquisition functions for Bayesian optimization
%U https://discovery.ucl.ac.uk/id/eprint/10083559/
%V 31
%X Bayesian optimization is a sample-efficient approach to global optimization that  relies on theoretically motivated value heuristics (acquisition functions) to guide  its search process. Fully maximizing acquisition functions produces the Bayes’  decision rule, but this ideal is difficult to achieve since these functions are frequently non-trivial to optimize. This statement is especially true when evaluating  queries in parallel, where acquisition functions are routinely non-convex, highdimensional, and intractable. We first show that acquisition functions estimated  via Monte Carlo integration are consistently amenable to gradient-based optimization. Subsequently, we identify a common family of acquisition functions, including EI and UCB, whose properties not only facilitate but justify use of greedy  approaches for their maximization.
%Z This version is the version of record. For information on re-use, please refer to the publisher’s terms and conditions.