Sequential Likelihood-Free Inference with Neural Proposal

TL;DR

This paper introduces Neural Proposal, a novel sampling method for likelihood-free Bayesian inference that ensures i.i.d. data collection, significantly improving the accuracy and efficiency of posterior estimation, especially for complex multi-modal distributions.

Contribution

The paper proposes Neural Proposal, a new sampling approach that addresses data degeneracy in sequential likelihood-free inference, ensuring unbiased, i.i.d. simulation inputs for better posterior estimation.

Findings

Improved posterior inference accuracy, especially for multi-modal distributions.

Significant reduction in simulation budget needed for reliable inference.

Enhanced performance over existing methods in various simulation scenarios.

Abstract

Bayesian inference without the likelihood evaluation, or likelihood-free inference, has been a key research topic in simulation studies for gaining quantitatively validated simulation models on real-world datasets. As the likelihood evaluation is inaccessible, previous papers train the amortized neural network to estimate the ground-truth posterior for the simulation of interest. Training the network and accumulating the dataset alternatively in a sequential manner could save the total simulation budget by orders of magnitude. In the data accumulation phase, the new simulation inputs are chosen within a portion of the total simulation budget to accumulate upon the collected dataset. This newly accumulated data degenerates because the set of simulation inputs is hardly mixed, and this degenerated data collection process ruins the posterior inference. This paper introduces a new sampling approach, called Neural Proposal (NP), of the simulation input that resolves the biased data collection as it guarantees the i.i.d. sampling. The experiments show the improved performance of our sampler, especially for the simulations with multi-modal posteriors.