Dataset of human-written and synthesized samples of keystroke dynamics features for free-text inputs.

The data presented in this article comprises human-written samples of keystroke dynamic features for free-text inputs, in the form of sentences written in natural language, together with synthesized samples that share the same text sequences. The human-written samples originate in three publicly available datasets that have been previously used in several keystroke dynamics studies; the corresponding synthesized samples, which have been forged as detailed in the companion article, share the same keystroke sequences as the human-written ones to facilitate comparison. The human-written samples were collected, and the synthesized samples created, with the objective of training and evaluating a liveness detection model. For each human-written sample of each source dataset and each method, 25 synthetic samples were included in the dataset here presented; these were forged using five different methods, a between-subject profile (only samples from users other than the target were available to the attacker) or with varying partial knowledge of the legitimate users' keystroke dynamics that ranged from only 100 keystrokes to all the available information. This dataset can be used by researchers to evaluate the performance of liveness detection methods for keystroke dynamics against a variety of state-of-the-art methods of sample synthesis.

On the shape of timings distributions in free-text keystroke dynamics profiles.

Keystroke dynamics is a soft biometric trait. Although the shape of the timing distributions in keystroke dynamics profiles is a central element for the accurate modeling of the behavioral patterns of the user, a simplified approach has been to presuppose normality. Careful consideration of the individual shapes for the timing models could lead to improvements in the error rates of current methods or possibly inspire new ones. The main objective of this study is to compare several heavy-tailed and positively skewed candidate distributions in order to rank them according to their merit for fitting timing histograms in keystroke dynamics profiles. Results are summarized in three ways: counting how many times each candidate distribution provides the best fit and ranking them in order of success, measuring average information content, and ranking candidate distributions according to the frequency of hypothesis rejection with an Anderson-Darling goodness of fit test. Seven distributions with two parameters and seven with three were evaluated against three publicly available free-text keystroke dynamics datasets. The results confirm the established use in the research community of the log-normal distribution, in its two- and three-parameter variations, as excellent choices for modeling the shape of timings histograms in keystroke dynamics profiles. However, the log-logistic distribution emerges as a clear winner among all two- and three-parameter candidates, consistently surpassing the log-normal and all the rest under the three evaluation criteria for both hold and flight times.