Neural Posterior Estimation with Latent Basis Expansions

ICLR 2026 Conference Submission21276 Authors

Keywords: forward KL divergence; simulation-based inference; variational inference; exponential family

TL;DR: Our approach parameterizes amortized variational distributions as linear combinations of basis functions; both the coefficients and basis functions themselves are fit by targeting the same variational objective.

Abstract:

Neural posterior estimation (NPE) is a likelihood-free amortized variational inference method that approximates projections of the posterior distribution. To date, NPE variational families have been either simple and interpretable (such as the Gaussian family) or highly flexible but black-box and potentially difficult to optimize (such as normalizing flows). In this work, we parameterize variational families via basis expansions of the latent variables. The log density of our variational distribution is a linear combination of latent basis functions (LBFs), which may be fixed a priori or adapted to the problem class of interest.
Our training and inference procedures are computationally efficient even for problems with high-dimensional latent spaces, provided only a low-dimensional projection of the posterior is of interest, owing to NPE's automatic marginalization capabilities.
In numerous inference problems, the proposed variational family exhibits better performance than existing variational families used with NPE, including mixtures of Gaussians (mixture density networks) and normalizing flows, as well as outperforming an existing basis expansion method for variational inference.

Primary Area: probabilistic methods (Bayesian methods, variational inference, sampling, UQ, etc.)

Submission Number: 21276