How to comprehensively describe a biometric update mechanisms for keystroke dynamics

Performance of biometric keystroke dynamics systems improve when using update mechanisms for the template. Generally, only the results are reported in literature but no description of how the results were achieved is presented. In this paper, we identify not only the parameters related to the update mechanism, but also the parameters for the testing of the update mechanism. For two selected update mechanisms, we will show that the performance of those mechanisms heavily depend on the used method for evaluation. For one of these update mechanisms the final FMR value varied from 12.85% to 53.60% depending on the evaluation method. This clearly shows that performance results on update mechanisms by themselves have no value unless the method of evaluation is described.