Oct 19 – 23, 2020
Europe/Zurich timezone

Black-Box Optimization with Local Generative Surrogates

Oct 21, 2020, 11:20 AM
Regular talk 10 ML for experimental particle physics Workshop


Mr Vladislav Belavin (Yandex School of Data Analysis (RU))


We propose a novel method for gradient-based optimization of black-box simulators using differentiable local surrogate models. In fields such as physics and engineering, many processes are modeled with non-differentiable simulators with intractable likelihoods. Optimization of these forward models is particularly challenging, especially when the simulator is stochastic. To address such cases, we introduce the use of deep generative models to iteratively approximate the simulator in local neighborhoods of the parameter space. We demonstrate that these local surrogates can be used to approximate the gradient of the simulator, and thus enable gradient-based optimization of simulator parameters. In cases where the dependence of the simulator on the parameter space is constrained to a low dimensional submanifold, we observe that our method attains minima faster than baseline methods, including Bayesian optimization, numerical optimization, and approaches using score function gradient estimators.

Primary authors

Mr Vladislav Belavin (Yandex School of Data Analysis (RU)) Sergey Shirobokov (Imperial College (GB)) Michael Aaron Kagan (SLAC National Accelerator Laboratory (US)) Andrey Ustyuzhanin (Yandex School of Data Analysis (RU)) Atılım Güneş Baydin (University of Oxford)

Presentation materials