Predictively-Oriented Kalman Filtering
Zheyang Shen, Gerardo Duran-Martin, Chris. J. Oates · 2026
Read Predictively-Oriented Kalman Filtering by Zheyang Shen, Gerardo Duran-Martin, Chris. J. Oates (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-03230.txt.
Description
Shen, Duran-Martin, and Oates build a Kalman-style filter that guards against wrong models. Online filtering tracks a hidden state from noisy data over time. If the dynamic or measurement model is off, classic Bayesian filters can grow over-confident. Their approach uses predictively oriented posteriors. In that frame, learning happens only when the whole model is well set.
Such posteriors are hard to characterize in closed form. The team's main tool is a fast update that mimics an iterated extended Kalman filter in a linear-Gaussian way. The method, EKF-PrO, has no tunable settings. Its cost is on par with standard filters. Tests cover linear and nonlinear cases where the state-space model is wrong on purpose. The filter avoids the over-confident errors that trap standard tools. It offers a principled hedge, not just a tweak to gain terms.
Catalog record
| CREATOR | Zheyang Shen, Gerardo Duran-Martin, Chris. J. Oates |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 55 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.03230