AI RESEARCH
From Gradients to Riccati Geometry: Kalman World Models for Single-Pass Learning
arXiv CS.LG
•
ArXi:2603.13423v1 Announce Type: new Backpropagation dominates modern machine learning, yet it is not the only principled method for optimizing dynamical systems. We propose Kalman World Models (KWM), a class of learned state-space models trained via recursive Bayesian filtering rather than reverse-mode automatic differentiation. Instead of gradient descent updates, we replace parameter learning with Kalman-style gain adaptation