Guaranteed Parameter Estimation for Cooperative Models

The parameters of cooperative models are estimated in a bounded-error context, i.e., all uncertain quantities are assumed to be bounded, with known bounds. Guaranteed estimation is then the characterization of the set of all parameter vectors that are consistent with the model and experimental data, given these bounds. Interval techniques provide an approximate but guaranteed enclosure of this set. No parameter vector consistent with the experimental data and model structure can be missed, so this approach bypasses the structural identifiability study required by the usual approaches based on the local optimization of some cost function.