A flexible state-space model for learning nonlinear dynamical systems
From MaRDI portal
Publication:2407186
DOI10.1016/J.AUTOMATICA.2017.02.030zbMATH Open1370.93300arXiv1603.05486OpenAlexW2963779771MaRDI QIDQ2407186FDOQ2407186
Authors: Andreas Svensson, Thomas B. Schön 
Publication date: 28 September 2017
Published in: Automatica (Search for Journal in Brave)
Abstract: We consider a nonlinear state-space model with the state transition and observation functions expressed as basis function expansions. The coefficients in the basis function expansions are learned from data. Using a connection to Gaussian processes we also develop priors on the coefficients, for tuning the model flexibility and to prevent overfitting to data, akin to a Gaussian process state-space model. The priors can alternatively be seen as a regularization, and helps the model in generalizing the data without sacrificing the richness offered by the basis function expansion. To learn the coefficients and other unknown parameters efficiently, we tailor an algorithm using state-of-the-art sequential Monte Carlo methods, which comes with theoretical guarantees on the learning. Our approach indicates promising results when evaluated on a classical benchmark as well as real data.
Full work available at URL: https://arxiv.org/abs/1603.05486
Recommendations
- Bayesian inference in nonparametric dynamic state-space models
- Online Bayesian inference and learning of Gaussian-process state-space models
- An Unsupervised Ensemble Learning Method for Nonlinear Dynamic State-Space Models
- Learning nonlinear state-space models using autoencoders
- Bayes factor estimation for nonlinear dynamic state space models
Gaussian processesregularizationnonlinear modelsMonte Carlo methodssystem identificationBayesian learningprobabilistic models
Nonlinear systems in control theory (93C10) Identification in stochastic control theory (93E12) Stochastic learning and adaptive control (93E35)
Cites Work
- System identification of nonlinear state-space models
- Title not available (Why is that?)
- Title not available (Why is that?)
- PRMLT
- Gaussian processes for machine learning.
- Title not available (Why is that?)
- Pattern recognition and machine learning.
- Title not available (Why is that?)
- Particle Markov Chain Monte Carlo Methods
- Convergence of a stochastic approximation version of the EM algorithm
- Some matrix-variate distribution theory: Notational considerations and a Bayesian application
- Bayesian system identification
- On the estimation of transfer functions, regularizations and Gaussian processes-revisited
- On particle methods for parameter estimation in state-space models
- Title not available (Why is that?)
- Prediction error identification of linear systems: a nonparametric Gaussian regression approach
- A new kernel-based approach for linear system identification
- Adaptive hinging hyperplanes and its applications in dynamic system identification
- Backward simulation methods for Monte Carlo statistical inference
- Nonlinear black-box modeling in system identification: A unified overview
- Robust maximum-likelihood estimation of multivariable dynamic systems
- Dynamic systems identification with Gaussian processes
- Nonlinear black-box models in system identification: Mathematical foundations
- Coupling a stochastic approximation version of EM with an MCMC procedure
- Neural networks for modelling and control of dynamic systems. A practitioner's handbook
- Nonlinear system identification via direct weight optimization
- Identification of nonlinear systems using polynomial nonlinear state space models
- Identification of dynamic systems using piecewise-affine basis function models
- Modelling and control of dynamic systems using Gaussian process models
- Recursive nonlinear-system identification using latent variables
- Unsupervised State-Space Modeling Using Reproducing Kernels
Cited In (20)
- Nonlinear system identification with regularized tensor network B-splines
- Riddled basins of the optimal states in learning dynamical systems
- Learning nonlinear state-space models using autoencoders
- Title not available (Why is that?)
- Dynamical system identification, model selection, and model uncertainty quantification by Bayesian inference
- Continuous-time system identification with neural networks: model structures and fitting criteria
- An Unsupervised Ensemble Learning Method for Nonlinear Dynamic State-Space Models
- Estimation of dual‐mode nonlinear stochastic systems with unknown parameters
- Modeling and identification of uncertain-input systems
- Unsupervised learning of observation functions in state space models by nonparametric moment methods
- A new method for sequential learning of states and parameters for state-space models: the particle swarm learning optimization
- Markov chain Monte Carlo based adaptive Rauch-Tung-Striebel smoother
- Input-to-state stability for system identification with continuous-time Runge–Kutta neural networks
- Reinforcement learning based MPC with neural dynamical models
- Recursive identification of a nonlinear state space model
- Highly computationally efficient state filter based on the delta operator
- Online Bayesian inference and learning of Gaussian-process state-space models
- Recursive estimation for sparse Gaussian process regression
- Collaborative linear dynamical system identification by scarce relevant/irrelevant observations
- Complex nonlinear dynamic system of oligopolies price game with heterogeneous players under noise
Uses Software
This page was built for publication: A flexible state-space model for learning nonlinear dynamical systems
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2407186)
