Adversarial Variational Optimization of Non-Differentiable Simulators
This addresses a key bottleneck in fields like science and engineering where complex simulators are used for generative modeling, offering a likelihood-free inference method for non-differentiable systems.
The paper tackled the problem of performing inference with non-differentiable simulators that lack tractable likelihoods by introducing Adversarial Variational Optimization (AVO), which learns a proposal distribution over parameters to minimize the Jensen-Shannon divergence between synthetic and observed data distributions.
Complex computer simulators are increasingly used across fields of science as generative models tying parameters of an underlying theory to experimental observations. Inference in this setup is often difficult, as simulators rarely admit a tractable density or likelihood function. We introduce Adversarial Variational Optimization (AVO), a likelihood-free inference algorithm for fitting a non-differentiable generative model incorporating ideas from generative adversarial networks, variational optimization and empirical Bayes. We adapt the training procedure of generative adversarial networks by replacing the differentiable generative network with a domain-specific simulator. We solve the resulting non-differentiable minimax problem by minimizing variational upper bounds of the two adversarial objectives. Effectively, the procedure results in learning a proposal distribution over simulator parameters, such that the JS divergence between the marginal distribution of the synthetic data and the empirical distribution of observed data is minimized. We evaluate and compare the method with simulators producing both discrete and continuous data.