Metraipolis › archive › dc_6f7cac024bf7

Gibbs Variational Last Layers for Long-Horizon Dynamics Learning — arXiv:2611.02891

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.02891
License (as archived)
CC0-1.0
Archived
2026-11-03T11:22:59.970Z
SHA-256 of the archived text
d889a2ba4091df0a60f25d71982d4f91c5cc8a6726ccd179a68dca2c5f2ff805
Stable URL of this page
https://metraipolis.com/documento/dc_6f7cac024bf7

How to cite

Gibbs Variational Last Layers for Long-Horizon Dynamics Learning — arXiv:2611.02891 — arXiv (en), https://arxiv.org/abs/2611.02891 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_6f7cac024bf7 (Metraipolis archive dc_6f7cac024bf7, sha256 d889a2ba4091df0a60f25d71982d4f91c5cc8a6726ccd179a68dca2c5f2ff805)
BibTeX
@misc{dc6f7cac024bf7,
  title = {Gibbs Variational Last Layers for Long-Horizon Dynamics Learning — arXiv:2611.02891},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.02891},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_6f7cac024bf7; sha256 d889a2ba4091df0a60f25d71982d4f91c5cc8a6726ccd179a68dca2c5f2ff805},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.02891. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Accurate long-horizon probabilistic predictions of dynamical systems are a prerequisite for model-based decision-making. Variational Bayesian last layers (VBLLs) are an attractive model class for this setting, offering tractable uncertainty at near-deterministic cost, yet they are trained exclusively with single-step likelihood objectives that may be insufficient for reliable multi-step rollouts. We revisit VBLLs through the lens of Gibbs variational inference, which decouples posterior inference from the choice of training loss. Building on this perspective, we introduce a family of multi-step training losses that progressively incorporate long-horizon structure, robustness via continuous ranked probability scores (CRPS), and training on full autoregressive rollouts. Our approach retains the simplicity of VBLLs while directly optimizing for multi-step predictive quality. Across illustrative synthetic environments, a diverse suite of chaotic dynamical systems, and real-world data, we show that multi-step, CRPS-based objectives substantially improve long-horizon accuracy and distributional fit, consistently outperforming single-step likelihood training.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG) ; Robotics (cs.RO).
Pagina del lavoro: https://arxiv.org/abs/2611.02891 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-11-03T11:02:37.655Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).