This contribution aims to enrich the recently introduced kernel-based regularization method for linear system identification. Instead of a single kernel, we use multiple kernels, which can be instances of any existing kernels for the impulse response estimation of linear systems. We also introduce a new class of kernels constructed based on output error (OE) model estimates. In this way, a more flexible and richer representation of the kernel is obtained. Due to this representation the associated hyper-parameter estimation problem has two good features. First, it is a difference of convex functions programming (DCP) problem. While it is still nonconvex, it can be transformed into a sequence of convex optimization problems with majorization minimization (MM) algorithms and a local minima can thus be found iteratively. Second, it leads to sparse hyper-parameters and thus sparse multiple kernels. This feature shows the kernel-based regularization method with multiple kernels has the potential to tackle various problems of finding sparse solutions in linear system identification.
|Title of host publication||2012 IEEE 51st Annual Conference on Decision and Control (CDC)|
|Publication status||Published - 2012|
|Event||51st IEEE Conference on Decision and Control - Maui, HI, United States|
Duration: 10 Dec 2012 → 13 Dec 2012
Conference number: 51
|Conference||51st IEEE Conference on Decision and Control|
|Period||10/12/2012 → 13/12/2012|