Robust EM kernel-based methods for linear system identification
From MaRDI portal
Abstract: Recent developments in system identification have brought attention to regularized kernel-based methods. This type of approach has been proven to compare favorably with classic parametric methods. However, current formulations are not robust with respect to outliers. In this paper, we introduce a novel method to robustify kernel-based system identification methods. To this end, we model the output measurement noise using random variables with heavy-tailed probability density functions (pdfs), focusing on the Laplacian and the Student's t distributions. Exploiting the representation of these pdfs as scale mixtures of Gaussians, we cast our system identification problem into a Gaussian process regression framework, which requires estimating a number of hyperparameters of the data size order. To overcome this difficulty, we design a new maximum a posteriori (MAP) estimator of the hyperparameters, and solve the related optimization problem with a novel iterative scheme based on the Expectation-Maximization (EM) method. In presence of outliers, tests on simulated data and on a real system show a substantial performance improvement compared to currently used kernel-based methods for linear system identification.
Recommendations
- A new kernel-based approach for linear system identification
- Kernels for linear time invariant system identification
- Nonparametric system identification by kernel methods
- Kernel-based identification of asymptotically stable continuous-time linear dynamical systems
- A New Kernel-Based Approach for NonlinearSystem Identification
- System identification using kernel-based regularization: new insights on stability and consistency issues
- Robust kernel clustering algorithm for nonlinear system identification
- Identification of non-linear systems by recursive kernel regression estimates
- On kernel design for regularized LTI system identification
Cites work
- scientific article; zbMATH DE number 45848 (Why is no real title available?)
- scientific article; zbMATH DE number 45871 (Why is no real title available?)
- scientific article; zbMATH DE number 65796 (Why is no real title available?)
- scientific article; zbMATH DE number 3551792 (Why is no real title available?)
- scientific article; zbMATH DE number 3567782 (Why is no real title available?)
- scientific article; zbMATH DE number 194744 (Why is no real title available?)
- scientific article; zbMATH DE number 3442988 (Why is no real title available?)
- scientific article; zbMATH DE number 840151 (Why is no real title available?)
- A new kernel-based approach for linear system identification
- A new look at the statistical model identification
- An $\ell _{1}$-Laplace Robust Kalman Smoother
- An Analysis of the EM Algorithm and Entropy-Like Proximal Point Methods
- Asymptotic properties of the least-squares method for estimating transfer functions and disturbance spectra
- Estimating the dimension of a model
- Gaussian processes for machine learning.
- Implementation of algorithms for tuning parameters in regularized least squares problems in system identification
- Kernel methods in system identification, machine learning and function estimation: a survey
- On covariance function tests used in system identification
- On the estimation of transfer functions, regularizations and Gaussian processes-revisited
- Prediction error identification of linear systems: a nonparametric Gaussian regression approach
- Regularization algorithms for learning that are equivalent to multilayer networks
- Regularized spectrum estimation using stable spline kernels
- Robust EM kernel-based methods for linear system identification
- Support Vector Method for RobustARMA System Identification
- The EM Algorithm and Extensions, 2E
Cited in
(20)- Kernel absolute summability is sufficient but not necessary for RKHS stability
- Learning linear modules in a dynamic network using regularized kernel-based methods
- A novel weight function-based robust iterative learning identification method for discrete Box-Jenkins models with Student's t-distribution noises
- Robust EM kernel-based methods for linear system identification
- High precision variational Bayesian inference of sparse linear networks
- On the mathematical foundations of stable RKHSs
- Regularized LTI system identification in the presence of outliers: a variational EM approach
- On semiseparable kernels and efficient implementation for regularized system identification and function estimation
- Bayesian positive system identification: truncated Gaussian prior and hyperparameter estimation
- A generalized minimal residual based iterative back propagation algorithm for polynomial nonlinear models
- Mercer kernel absolute integrability is only sufficient for RKHS stability
- Articulated estimator random field and geometrical approach applied in system identification
- An empirical Bayes approach to identification of modules in dynamic networks
- A new kernel-based approach to system identification with quantized output data
- Global convergence of the EM algorithm for ARX models with uncertain communication channels
- Adaptive regularised kernel-based identification method for large-scale systems with unknown order
- Augmented flexible least squares algorithm for time‐varying parameter systems
- Robust identification of linear ARX models with recursive EM algorithm based on Student's t-distribution
- Some stochastic gradient algorithms for Hammerstein systems with piecewise linearity
- Boosting as a kernel-based method
This page was built for publication: Robust EM kernel-based methods for linear system identification
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q259415)
