Salvato in:
| Autori principali: | , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2510.03614 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866912629328445440 |
|---|---|
| author | Solinas, Christopher Haluska, Radovan Sychrovsky, David Timbers, Finbarr Bard, Nolan Buro, Michael Schmid, Martin Sturtevant, Nathan R. Bowling, Michael |
| author_facet | Solinas, Christopher Haluska, Radovan Sychrovsky, David Timbers, Finbarr Bard, Nolan Buro, Michael Schmid, Martin Sturtevant, Nathan R. Bowling, Michael |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_03614 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Neural Bayesian Filtering Solinas, Christopher Haluska, Radovan Sychrovsky, David Timbers, Finbarr Bard, Nolan Buro, Michael Schmid, Martin Sturtevant, Nathan R. Bowling, Michael Machine Learning Artificial Intelligence 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. |
| title | Neural Bayesian Filtering |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2510.03614 |