A review on the public benchmark databases for static keystroke dynamics

Abstract

Keystroke dynamics allows to authenticate individuals through their way of typing their password or a free text on a keyboard. In general, in biometrics, a novel algorithm is validated through a comparison to the state of the art one's using some datasets in an offline way. Several benchmark datasets for keystroke dynamics have been proposed in the literature. They differ in many ways and their intrinsic properties influence the performance of the algorithms under evaluation. In this work, we (a) provide a literature review on existing benchmark datasets of keystroke dynamics; (b) present several criteria and tests in order to characterize them; (c) and apply these criteria on these available public benchmark datasets. The review analysis shows a great disparity in the acquisition protocol, the population involved, the complexity of the passwords, or the expected performance (there is a relative difference of 76% between the EER on the worst and best performing datasets with the same authentication method).

Analysis of Keystroke Dynamics for the Generation of Synthetic Datasets

“…Indeed, they work in an offline context by using samples previously collected by other researchers, and stored in a benchmark dataset. A complete list of available keystroke dynamics datasets has been made in [4,5]. As it can be seen, most of datasets have less than 200 individuals and few samples for each user.…”

Statistical modeling of keystroke dynamics samples for the generation of synthetic datasets

“…First, diagonals of correlation matrix are discarded. Correlations between two durations DT Di [5], and DT Dj As shown in Figure 4, no strong stable correlation has been found between durations from different Digraph, (Out: dataset, Out U: dataset splitted by User). DigraphTime will be thus assumed independent.…”

“…Indeed, they work in an offline context by using samples previously collected by other researchers, and stored in a benchmark dataset. A complete list of available keystroke dynamics datasets has been made in [4,5]. As it can be seen, most of datasets have less than 200 individuals and few samples are available for each user.…”

Zoo Graph: A New Visualisation for Biometric System Evaluation

“…Among these reasons, we can list: the implementation of biometric authentication systems relies on machine learning and pattern recognition systems that are prone for errors, the biometric data can evolve over time, the biometric sensor can be noisy, the biometric modality is not universal, etc. For this reason, many researchers focused on developing performance evaluation frameworks to evaluate the overall performance of biometric systems [2]. This is done by developing dedicated biometric databases and performance metrics to be used to evaluate and compare new authentication algorithms.…”