A one-step-per-layer smoother that approximates as Gaussian only each layer's affine output is derived that applies both to deterministic systems with noisy observations and to stochastic systems described by output statistics.
Abstract
A neural network's layers can be treated as time steps of a state-space model, turning Bayesian training into a smoothing problem: a forward pass propagates Gaussian moments through the network, and a backward Rauch--Tung--Striebel pass updates the weight posteriors in closed form. Such methods learn from each observation in a single pass, in an uncertainty-aware manner, and without gradient-based iterations or replay, which makes them well suited for online adaptation and data-efficient learning. Existing smoothing-based methods, however, are restricted to diagonal covariances across activations, discarding correlations between neurons. We overcome this limitation via a cross-covariance identity that enables full-covariance propagation through a network's nonlinear activations. We derive a one-step-per-layer smoother that approximates as Gaussian only each layer's affine output, and that applies both to deterministic systems with noisy observations and to stochastic systems described by output statistics. We demonstrate this method in non-stationary classification, online dynamics learning, and policy adaptation of a vision-language-action model, and find that it is generally more accurate than other smoothing-based methods.
It is demonstrated that performance can be significantly improved by retaining all data and using a learning model that reflects the temporal structure of the environment.
Matt Jones, David Mayo, Tyler R. Scott et al.· 3 citations
The modified Bryson-Frazier (MBF) smoother is revisited as an alternative to the RTS smoother for Gaussian process regression in its state-space representation and it is demonstrated that the intermediate quantities computed by the MBF smoother can be reused to compute gradients of the negative log marginal likelihood,...
It is demonstrated that the posterior density can be written as a mixture density with suitably defined auxiliary random variables, where the mixture components are log-concave, and neuron parameters can be sampled from the posterior by only sampling log-concave densities.
Curtis McDonald, Andrew R. Barron· Mathematical Statistics and...· 1 citation
Analytical and particle-based methods for uncertainty propagation in random neural network models, where both the inputs and network parameters are allowed to be random, are developed and extended to autonomous dynamical systems whose one-step evolution map is represented by a random neural network.
A Bayesian, online DPC framework based on autoregressive models with exogenous inputs (ARX) that uses an externally-provided prior, which encodes inductive bias such as smooth system dynamics and stability, to safeguard performance when SNR is low is proposed.
Alessandro Chiuso, F. Dörfler, Keith Moffat· 0 citations
This work develops a framework for source-space generalized Bayesian inference that combines inexpensive few-step prior transports with posterior stability guarantees and establishes Wasserstein error bounds between the exact and learned posteriors in terms of the joint population iMF and auxiliary-velocity loss.
Hoang Phuc Hau Luu, Marcelo Hartmann, Zhong-Jian Wang· 0 citations
Related blog posts
MIT News · Artificial Intelligence· news.mit.eduNov 17, 2026
By studying the computational foundations of language learning, Associate Professor Jacob Andreas builds intelligent systems that can communicate effectively with humans.
Microsoft Research Blog· microsoft.comNov 16, 2026
AI coding agents are increasingly help developers fix bugs and build features largely on their own, but every interaction with their models consumes tokens. In each session, an agent sends instructions, tool descriptions, and session history to its model. Most efforts to reduce this cost focus on individual tasks. Yet developers return to the same […] The post TokenEconomizer: Learning from Experience to Make AI Coding Agents More Efficient appeared first on Microsoft Research.
MIT News · Artificial Intelligence· news.mit.eduNov 13, 2026