Black-Box Optimization with Local Generative Surrogates
This addresses optimization challenges in fields like physics and engineering where simulators are stochastic and intractable, offering a novel method for a specific bottleneck.
The paper tackles the problem of gradient-based optimization for non-differentiable black-box simulators by using deep generative models as local surrogates to approximate gradients, enabling faster convergence to minima compared to baseline methods like Bayesian optimization.
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.