Abstract:Data assimilation methods aim at estimating the state of a system by combining observations with a physical model. When sequential data assimilation is considered, the joint distribution of the latent state and the observations is described mathematically using a state-space model, and filtering or smoothing algorithms are used to approximate the conditional distribution of the state given the observations. The most popular algorithms in the data assimilation community are based on the Ensemble Kalman Filter and Smoother (EnKF/EnKS) and its extensions. In this paper we investigate an alternative approach where a Conditional Particle Filter (CPF) is combined with Backward Simulation (BS). This allows to explore efficiently the latent space and simulate quickly relevant trajectories of the state conditionally to the observations. We also tackle the difficult problem of parameter estimation. Indeed, the models generally involve statistical parameters in the physical models and/or in the stochastic models for the errors. These parameters strongly impact the results of the data assimilation algorithm and there is a need for an efficient method to estimate them. Expectation-Maximization (EM) is the most classical algorithm in the statistical literature to estimate the parameters in models with latent variables. It consists in updating sequentially the parameters by maximizing a likelihood function where the state is approximated using a smoothing algorithm. In this paper, we propose an original Stochastic Expectation-Maximization (SEM) algorithm combined to the CPF-BS smoother to estimate the statistical parameters. We show on several toy models that this algorithm provides, with reasonable computational cost, accurate estimations of the statistical parameters and the state in highly nonlinear state-space models, where the application of EM algorithms using EnKS is limited. We also provide a Python source code of the algorithm.

Likelihood-free stochastic approximation EM for inference in complex models

An easy-to-use empirical likelihood ABC method

Affinely Parametrized State-space Models: Ways to Maximize the Likelihood Function

An efficient particle-based method for maximum likelihood estimation in nonlinear state-space models

When Composite Likelihood Meets Stochastic Approximation

Quasi Maximum Likelihood Estimation and Inference of Large Approximate Dynamic Factor Models via the EM algorithm

Stochastic Approximation with Biased MCMC for Expectation Maximization

Maximum Likelihood from Incomplete Data Via the EM Algorithm

On an Extension of Stochastic Approximation EM Algorithm for Incomplete Data Problems

Model uncertainty estimation using the expectation maximization algorithm and a particle flow filter

Particle Gibbs for Likelihood-Free Inference of State Space Models with Application to Stochastic Volatility

An adaptive simulated annealing EM algorithm for inference on non-homogeneous hidden Markov models

Model error covariance estimation in particle and ensemble Kalman filters using an online expectation-maximization algorithm

Performance in population models for count data, part II: a new SAEM algorithm

Bayesian synthetic likelihood for stochastic models with applications in mathematical finance

A smoothed semiparametric likelihood for estimation of nonparametric finite mixture models with a copula-based dependence structure

Vecchia Likelihood Approximation for Accurate and Fast Inference in Intractable Spatial Extremes Models

Maximum likelihood estimation via the ECM algorithm: A general framework

Efficient stochastic optimisation by unadjusted Langevin Monte Carlo. Application to maximum marginal likelihood and empirical Bayesian estimation

Stochastic Expectation Maximization with Variance Reduction.

Deterministic Approximate EM Algorithm; Application to the Riemann Approximation EM and the Tempered EM