A wrapper based feature subset selection using ACO-ELM-ANP and GA-ELM-ANP approaches for keystroke dynamics authentication

The security of computer access is important today because of huge transactions being carried out every day via the Internet. Username with password is the commonly used authentication mechanism. Most of the text based authentication methods are vulnerable to many attacks as they depend on text and can be strengthened more by combining password with key typing manner of the user. Keystroke Dynamics is one of the famous and inexpensive behavioral biometric technologies, which identifies the authenticity of a user when the user is working via a keyboard. The paper uses a new feature called Virtual Key Force along with the commonly extracted timing features. Features are normalized using Z-Score method. For feature subset selection, wrapper based approach using Ant Colony Optimization - Extreme Learning Machine with Analytic Network Process (ACO-ELM-ANP) and Genetic Algorithm - Extreme Learning Machine with Analytic Network Process (GA-ELM-ANP) are proposed. From the results, it is observed that ACO-ELM-ANP selects less number of features for further processing.