|The BANCA Protocol|
An evaluation protocol defines a set of data, how it should be used by a system to perform a set of experiments and how the system performance should be computed.
In verification, two types of protocols exist; closed-set and open-set. In closed-set verification the population of clients is fixed. This means that the system design can be tuned to the clients in the set. Thus both the adopted representation (features) and the verification algorithm applied in the feature space are based on some training data collected for this set of clients. Anyone who is not in the training set is considered an impostor. The XM2VTS protocol is an example of this type of verification problem formulation.
In open-set verification we wish to add new clients to the list without having to redesign the verification system. In particular, we want to use the same feature space and the same design parameters such as thresholds. In such a scenario the feature space and the verification system parameters must be trained using completely independent data from that used for specifying client models. The BANCA protocol is an example of an open-set verification protocol.