Robust EM kernel-based methods for linear system identification. (May 2016)

• Record Type: Journal Article
• Authors: Bottegal, Giulio; Aravkin, Aleksandr Y.; Hjalmarsson, Håkan; Pillonetto, Gianluigi
• Is Part Of: Automatica. Volume 67(2016)
• Page Start: 114
• Page End: 126
• Publication Date: 2016-05
• Subjects: System identification -- Kernel-based methods -- Outliers -- MAP estimate -- EM method; Automatic control -- Periodicals; Automation -- Periodicals; 629.805
• DOI: 10.1016/j.automatica.2016.01.036
• Languages: English
• ISSNs: 0005-1098

Abstract

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 the 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.