Место издания:IEEE Computer Society [New York], United States
Первая страница:61
Последняя страница:66
Аннотация:Nowadays there is a growing interest to activeauthentication methods in security society. These methods areused for user identity validation with behavioral biometricssuch as keystroke or mouse moving dynamics. In this article anew hybrid method for active authentication using keystrokedynamics is presented. This method is a combination of a newkeystroke data representation model based on potentialfunctions and machine learning algorithms based on decisiontrees. The proposed method is tested in static and dynamicauthentication scenarios on the benchmark Si6 dataset anddatasets collected by the authors. The experimental resultsconfirm that the proposed hybrid method is applicable to reallifeauthentication systems.