Novel approximations for inference in nonlinear dynamical systems using expectation propagation

Authors:
Alexander Ypma;Tom Heskes
Affiliations:
SNN, Radboud University Nijmegen, Geert Grooteplein 21, 6525 EZ Nijmegen, The Netherlands;SNN, Radboud University Nijmegen, Geert Grooteplein 21, 6525 EZ Nijmegen, The Netherlands
Venue:
Neurocomputing
Year:
2005

Citing 7
Cited 4

Numerical Recipes in C: The Art of Scientific Computing

Numerical Recipes in C: The Art of Scientific Computing
An unsupervised ensemble learning method for nonlinear dynamic state-space models

Neural Computation
A family of algorithms for approximate bayesian inference

A family of algorithms for approximate bayesian inference
Information Theory, Inference & Learning Algorithms

Information Theory, Inference & Learning Algorithms
Expectation propagation for approximate inference in dynamic bayesian networks

UAI'02 Proceedings of the Eighteenth conference on Uncertainty in artificial intelligence
Monitoring a complex physical system using a hybrid dynamic bayes net

UAI'02 Proceedings of the Eighteenth conference on Uncertainty in artificial intelligence
Factor graphs and the sum-product algorithm

IEEE Transactions on Information Theory

Neural Decoding of Movements: From Linear to Nonlinear Trajectory Models

Neural Information Processing
EP for Efficient Stochastic Control with Obstacles

Proceedings of the 2010 conference on ECAI 2010: 19th European Conference on Artificial Intelligence
Efficient computation of the maximum a posteriori path and parameter estimation in integrate-and-fire and more general state-space models

Journal of Computational Neuroscience
Sparse Spatio-temporal Gaussian processes with general likelihoods

ICANN'11 Proceedings of the 21th international conference on Artificial neural networks - Volume Part I

Quantified Score

Hi-index	0.01

Visualization

Abstract

We formulate the problem of inference in nonlinear dynamical systems in the framework of expectation propagation, and propose two novel algorithms. The first algorithm is based on Laplace approximation and allows for iterated forward and backward passes. The second is based on repeated application of the unscented transform. It leads to an unscented Kalman smoother for which the dynamics need not be inverted explicitly. In experiments with a one-dimensional nonlinear dynamical system we show that for relatively low observation noise levels, the Laplace algorithm allows for the best estimates of the state means. The unscented algorithm however is more robust to high observation noise and always outperforms the conventional inference methods against which it was benchmarked.