Bayesian Optimization with a Prior for the Optimum

While Bayesian Optimization (BO) is a very popular method for optimizing expensive black-box functions, it fails to leverage the experience of domain experts. This causes BO waste function evaluations on bad design choices (e.g., machine learning hyperparameters) that expert already knows work poorly. To address this issue, we introduce with Prior Optimum (BOPrO). BOPrO allows users inject their knowledge into optimization process in form priors about which parts input space will yield best performance, rather than BO’s standard over are much less intuitive users. then combines these probabilistic model pseudo-posterior used select points evaluate next. We show around \(6.67\times \) faster state-of-the-art methods common suite benchmarks, and achieves new performance real-world hardware application. also converges even if optimum not entirely accurate robustly recovers from misleading priors.