Bayesian Inference and Learning in Gaussian Process State-Space Models with Particle MCMC

Abstract

State-space models are successfully used in many areas of science, engineering and economics to model time series and dynamical systems. We present a fully Bayesian approach to inference \emph{and learning} (i.e. state estimation and system identification) in nonlinear nonparametric state-space models. We place a Gaussian process prior over the state transition dynamics, resulting in a flexible model able to capture complex dynamical phenomena. To enable efficient inference, we marginalize over the transition dynamics function and infer directly the joint smoothing distribution using specially tailored Particle Markov Chain Monte Carlo samplers. Once a sample from the smoothing distribution is computed, the state transition predictive distribution can be formulated analytically. Our approach preserves the full nonparametric expressivity of the model and can make use of sparse Gaussian processes to greatly reduce computational complexity.

References in corpus (1)

• Ancestor Sampling for Particle Gibbs

Cited by in corpus (17)

• When Gaussian Process Meets Big Data: A Review of Scalable GPs
• Continuous time Gaussian process dynamical models in gene regulatory network inference
• Considering discrepancy when calibrating a mechanistic electrophysiology model
• Probabilistic learning of nonlinear dynamical systems using sequential Monte Carlo
• Gaussian Process Volatility Model
• Online Joint State Inference and Learning of Partially Unknown State-Space Models
• An Introduction to Gaussian Process Models
• FedLoc: Federated Learning Framework for Data-Driven Cooperative Localization and Location Data Processing
• Identification of Gaussian Process State Space Models
• Hybrid Gaussian Process Modeling Applied to Economic Stochastic Model Predictive Control of Batch Processes
• Recursive Estimation for Sparse Gaussian Process Regression
• Learning Nonlinear State Space Models with Hamiltonian Sequential Monte Carlo Sampler
• Bayesian Hidden Physics Models: Uncertainty Quantification for Discovery of Nonlinear Partial Differential Operators from Data
• A projected nonlinear state-space model for forecasting time series signals
• Learning-Based Optimal Control with Performance Guarantees for Unknown Systems with Latent States
• The Renyi Gaussian Process: Towards Improved Generalization
• Prediction with Approximated Gaussian Process Dynamical Models