Self adaptive systems: an experimental analysis of the performance over time

Recently adaptive biometric systems have received significant boost in the research community. These systems have the ability to automatically adapt/ update templates to the variation of input samples in the changing environment. Among all, self adaptive systems have gained substancial interest. These systems adapt themselves to the changing environment using highly confidently classified input samples available during the system's operation. However, although being commonly accepted that self adaptive biometric systems may be robust against short and medium-term, temporal, lighting and expression variations, no paper has shown its robustness over time neither theoretically nor experimentally. Moreover, evidences of performance assessment are usually drawn on small sample sized database collected over short term. This paper evaluates the performance of self adaptive systems over time on DIEE multi-modal face and fingerprint dataset, explicitly collected to this aim. Experimental validations prove that self adapting results in robustness of the classifier over time.