Black Box Optimization via a Bayesian-Optimized Genetic Algorithm
Abstract
We present a simple and robust optimization algorithm related to genetic algorithms, and with analogies to the popular CMA-ES search algorithm, that serves as a cheap alternative to Bayesian Optimization. The algorithm is robust against both monotonic transforms of the objective function value and affine transformations of the feasible region. It is fast and easy to implement, and has performance comparable to CMA-ES on a suite of benchmarks while spending less CPU in the optimization algorithm, and can exhibit better overall performance than Bayesian Optimization when the objective function is cheap.