Approximate Bayesian Computation with Domain Expert in the Loop
This work addresses a bottleneck in likelihood-free inference for researchers and practitioners using ABC, offering an incremental improvement by automating statistic selection with expert input.
The paper tackles the laborious task of selecting summary statistics in Approximate Bayesian Computation (ABC) by introducing an active learning method that involves domain experts, reducing their workload and handling misspecified models. Empirical results show improved posterior estimates compared to existing methods under limited simulation budgets.
Approximate Bayesian computation (ABC) is a popular likelihood-free inference method for models with intractable likelihood functions. As ABC methods usually rely on comparing summary statistics of observed and simulated data, the choice of the statistics is crucial. This choice involves a trade-off between loss of information and dimensionality reduction, and is often determined based on domain knowledge. However, handcrafting and selecting suitable statistics is a laborious task involving multiple trial-and-error steps. In this work, we introduce an active learning method for ABC statistics selection which reduces the domain expert's work considerably. By involving the experts, we are able to handle misspecified models, unlike the existing dimension reduction methods. Moreover, empirical results show better posterior estimates than with existing methods, when the simulation budget is limited.