Segmentation of ARX models through SDP-relaxation techniques

Segmentation of ARX models can be formulated as a combinatorial minimization problem in terms of the l 0 -norm of the parameter variations and the l 2 -loss of the prediction error. A typical approach to compute an approximate solution to such a problem is based on l 1 -relaxation. Unfortunately, evaluation of the level of accuracy of the l 1 -relaxation in approximating the optimal solution of the original combinatorial problem is not easy to accomplish. In this poster, an alternative approach is proposed which provides an attractive solution for the l 0 -norm minimization problem associated with segmentation of ARX models.