Conditional diffusions for amortized neural posterior estimation
This work addresses simulation-based Bayesian inference for researchers, offering incremental improvements over prior diffusion-based NPE methods.
The paper tackled the limitations of flow-based neural posterior estimation (NPE) methods, such as training instability and computational trade-offs, by proposing conditional diffusions with summary networks, resulting in improved stability, superior accuracy, and faster training times across benchmarking problems.
Neural posterior estimation (NPE), a simulation-based computational approach for Bayesian inference, has shown great success in approximating complex posterior distributions. Existing NPE methods typically rely on normalizing flows, which approximate a distribution by composing many simple, invertible transformations. But flow-based models, while state of the art for NPE, are known to suffer from several limitations, including training instability and sharp trade-offs between representational power and computational cost. In this work, we demonstrate the effectiveness of conditional diffusions coupled with high-capacity summary networks for amortized NPE. Conditional diffusions address many of the challenges faced by flow-based methods. Our results show that, across a highly varied suite of benchmarking problems for NPE architectures, diffusions offer improved stability, superior accuracy, and faster training times, even with simpler, shallower models. Building on prior work on diffusions for NPE, we show that these gains persist across a variety of different summary network architectures. Code is available at https://github.com/TianyuCodings/cDiff.