Performance Comparison of Particle Swarm Optimization and Genetic Algorithm for Feature Subset Selection in Keystroke Dynamics

摘要

In recent years there has been increased focus on use of keystroke dynamics based authentication for mobile phones. The accuracy of the keystroke dynamics usually increases with an increase in features. But due to the limited processing power of mobile phones, it becomes essential to keep the feature subset minimal and at the same time keeping the accuracy unaffected. Thus, optimization algorithms play a vital role for feature subset selection. We compared the performance of Particle Swarm Optimization and Genetic Algorithm in reducing the number of features. It was observed that the feature set reduced from 49 to a range between 17-26 and the accuracy of the system increased to a maximum of 92.58%.