ViBR-WM: Visual Bayesian Regression for World Modeling
Abstract
Modeling temporal dependence and uncertainty is central to forecasting with world models. The Visual Bayesian Regression World Model combines visual features, physical histories and known covariates through interpretable regression, within a modular architecture supporting trend, seasonal and cycle dynamics. Visual compression reduces representation dimension, while Bayesian variable selection reduces active regression dimension. Posterior prediction combines forecasts across predictor subsets using their posterior probabilities as weights and accounts for parameter uncertainty and future disturbances. The model forecasts joint visual--physical states recursively and physical targets directly. Across four forecasting tasks spanning object motion, vegetation greenness and solar power, ViBR-WM achieves lower mean overall physical-target error than Temporal Straightening, ConvLSTM, PredRNN and SimVP on every task. Repeated fitting and resampling support these overall gains.
Related
- Kumaraswamy-Teissier Distribution: Bayesian and Non-Bayesian Inference, Regression Modeling and Climate Application
- BGM-IV: AI-Powered Bayesian Generative Modeling for Instrumental Variable Regression with High-Dimensional Covariates
- Modeling bounded well-being indices using Bayesian double generalized beta regression with spatial and temporal borrowing
- Online Bayesian Model Averaging with Joint Uncertainty Quantification for Models and Regression Coefficients in Binary Regression
- Recursive moment updates for scalable Bayesian nonparametric regression
BibTeX
@misc{li2026vibrwm,
title = {{ViBR-WM: Visual Bayesian Regression for World Modeling}},
author = {Jifan Li and Ning Ning},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.33844},
eprint = {2609.33844},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.33844},
}
RIS
TY - UNPB
TI - ViBR-WM: Visual Bayesian Regression for World Modeling
AU - Jifan Li
AU - Ning Ning
PY - 2026
DA - 2026/09/27
PB - arXiv
DO - 10.48550/arxiv.2609.33844
UR - https://arxiv.org/abs/2609.33844
AB - Modeling temporal dependence and uncertainty is central to forecasting with world models. The Visual Bayesian Regression World Model combines visual features, physical histories and known covariates through interpretable regression, within a modular architecture supporting trend, seasonal and cycle dynamics. Visual compression reduces representation dimension, while Bayesian variable selection reduces active regression dimension. Posterior prediction combines forecasts across predictor subsets using their posterior probabilities as weights and accounts for parameter uncertainty and future disturbances. The model forecasts joint visual--physical states recursively and physical targets directly. Across four forecasting tasks spanning object motion, vegetation greenness and solar power, ViBR-WM achieves lower mean overall physical-target error than Temporal Straightening, ConvLSTM, PredRNN and SimVP on every task. Repeated fitting and resampling support these overall gains.
ER -