Neural Bayesian Filtering
Abstract
We present Neural Bayesian Filtering (NBF), an algorithm for maintaining distributions over hidden states, called beliefs, in partially observable systems. NBF is trained to find a good latent representation of the beliefs induced by a task. It maps beliefs to fixed-length embedding vectors, which condition generative models for sampling. During filtering, particle-style updates compute posteriors in this embedding space using incoming observations and the environment's dynamics. NBF combines the computational efficiency of classical filters with the expressiveness of deep generative models - tracking rapidly shifting, multimodal beliefs while mitigating the risk of particle impoverishment. We validate NBF in state estimation tasks in three partially observable environments.
Cite
@article{arxiv.2510.03614,
title = {Neural Bayesian Filtering},
author = {Christopher Solinas and Radovan Haluska and David Sychrovsky and Finbarr Timbers and Nolan Bard and Michael Buro and Martin Schmid and Nathan R. Sturtevant and Michael Bowling},
journal= {arXiv preprint arXiv:2510.03614},
year = {2025}
}