Using Maximum Likelihood (or Prediction Error) methods to identify linear state space model is a prime technique. The likelihood function is a nonconvex function and care must be exercised in the numerical maximization. Here the focus will be on affine parameterizations which allow some special techniques and algorithms. Three approaches to formulate and perform the maximization are described in this contribution: (1) The standard and well known Gauss-Newton iterative search, (2) a scheme based on the EM (expectation-maximization) technique, which becomes especially simple in the affine parameterization case, and (3) a new approach based on lifting the problem to a higher dimension in the parameter space and introducing rank constraints.
|Publication status||Published - 2018|
|Event||SYSID 2018: 18th IFAC Symposium on System Identification - Stockholm, Sweden|
Duration: 9 Jul 2018 → 11 Jul 2018
- difference-of-convex optimization
- expectation-maximization algorithm
- maximum-likelihood estimation
- Parameterized state-space model