BayesFlow: Learning complex stochastic models with invertible neural networks
This provides a general framework for amortized Bayesian parameter estimation in scientific fields where traditional inference methods fail due to model complexity, though it builds on existing neural network techniques.
The authors tackled the problem of estimating parameters for complex stochastic models without explicit likelihood functions by proposing BayesFlow, a method using invertible neural networks for globally amortized Bayesian inference, which achieved efficient posterior inference on challenging intractable models from various domains.
Estimating the parameters of mathematical models is a common problem in almost all branches of science. However, this problem can prove notably difficult when processes and model descriptions become increasingly complex and an explicit likelihood function is not available. With this work, we propose a novel method for globally amortized Bayesian inference based on invertible neural networks which we call BayesFlow. The method uses simulation to learn a global estimator for the probabilistic mapping from observed data to underlying model parameters. A neural network pre-trained in this way can then, without additional training or optimization, infer full posteriors on arbitrary many real datasets involving the same model family. In addition, our method incorporates a summary network trained to embed the observed data into maximally informative summary statistics. Learning summary statistics from data makes the method applicable to modeling scenarios where standard inference techniques with hand-crafted summary statistics fail. We demonstrate the utility of BayesFlow on challenging intractable models from population dynamics, epidemiology, cognitive science and ecology. We argue that BayesFlow provides a general framework for building amortized Bayesian parameter estimation machines for any forward model from which data can be simulated.