Differentiating the Black-Box: Optimization with Local Generative Surrogates

by   Sergey Shirobokov, et al.

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 all baseline methods, including Bayesian optimization, numerical optimization, and REINFORCE driven approaches.


page 8

page 14

page 15

page 16

page 17


Zero Grads Ever Given: Learning Local Surrogate Losses for Non-Differentiable Graphics

Gradient-based optimization is now ubiquitous across graphics, but unfor...

Generative Evolutionary Strategy For Black-Box Optimizations

Many scientific and technological problems are related to optimization. ...

DiffTune: Optimizing CPU Simulator Parameters with Learned Differentiable Surrogates

CPU simulators are useful tools for modeling CPU execution behavior. How...

Enabling Simulation-Based Optimization Through Machine Learning: A Case Study on Antenna Design

Complex phenomena are generally modeled with sophisticated simulators th...

Branches of a Tree: Taking Derivatives of Programs with Discrete and Branching Randomness in High Energy Physics

We propose to apply several gradient estimation techniques to enable the...

Differentiable Multi-Target Causal Bayesian Experimental Design

We introduce a gradient-based approach for the problem of Bayesian optim...

Code Repositories


Local Generative Surrogate Optimisation algorithm

view repo

Please sign up or login with your details

Forgot password? Click here to reset