Premium
Realtime sequential inference of static parameters with expensive likelihood calculations
Author(s) -
Robins P.,
Rapley V. E.,
Green N.
Publication year - 2009
Publication title -
journal of the royal statistical society: series c (applied statistics)
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.205
H-Index - 72
eISSN - 1467-9876
pISSN - 0035-9254
DOI - 10.1111/j.1467-9876.2009.00670.x
Subject(s) - inference , markov chain monte carlo , computer science , monte carlo method , particle filter , algorithm , importance sampling , sampling (signal processing) , approximate inference , artificial intelligence , mathematics , kalman filter , statistics , bayesian probability , filter (signal processing) , computer vision
Summary. A methodology is developed for making inference about parameters of a possible covert chemical or biological atmospheric release from sensor readings. The key difficulty in performing this inference is that the results must be obtained in a very short timescale (5 min) to make use of the inference for protection. The methodology that is developed uses some of the components in a sequential Monte Carlo algorithm. However, this inference problem is different from many other sequential Monte Carlo problems, in that there are no state evolution equations, the forward model is highly non‐linear and the likelihoods are non‐Gaussian. The algorithm that is developed can use stored output from complex physics models for more rapid update of the posterior from new data without having to rerun the models. The use of differential evolution Markov chain sampling allows new samples to diverge rapidly from degenerate sample sets. Results for inferences made of atmospheric releases (both real and simulated) of material are presented, demonstrating that the sampling scheme performs adequately despite constraints of a short time span for calculations.