Quasi-OBE set estimation in parametric classification problems

Emerging set-theoretic identification and filtering techniques provide interesting new approaches to solving classification problems involving parametric structures. Of this class of methods, the recently developed quasi-OBE (QOBE) algorithm is particularly useful in real-time signal processing applications involving signal and system classification. QOBE estimation yields a useful set solution, while exhibiting innovative use of observation data, excellent convergence rates, and superb tracking performance. This paper explores strategies for the use of QOBE set solutions in classification problems. By exploiting the properties of the algorithm that are amenable to such solutions, classification performance is significantly improved with respect to conventional approaches