Recent Articles and Multimedia

Paper Title: Tempering the Bayes Filter towards Improved Model-Based Estimation

Authors: Menno van Zutphen, Domagoj Herceg, Giannis Delimpaltadakis, and Duarte J. Antunes

Model-based filtering is often performed under imperfect system models due, for example, to limited data availability for system identification, and recent work in Bayesian inference suggests that distributional tempering can improve predictive accuracy. In this paper, the authors develop the discrete-time tempered Bayes filter — and, in the linear-Gaussian case, a tempered Kalman filter — that improves belief-distribution estimates through multiple tempering modalities while preserving the recursivity and computational complexity of the original untempered filters.

READ MORE ABOUT THIS PAPER HERE