Improving feature extraction in keystroke dynamics using Optimization Techniques and Neural Network

This paper presents a novel application of optimization technique to user identity authentication using keystroke dynamics. Keystroke dynamics is a biometric technique to identify a user based on the analysis of his/her typing rhythm. Mean, Median and Standard deviation of feature values such as Latency, Duration and Digraph are measured and compared the performance. Particle Swarm Optimization (PSO) and Genetic Algorithm (GA) are used to select the subset of the features extracted and Neural Net is used for classification. Particle Swarm Optimization gives moderate performance than Genetic Algorithm with regard to feature reduction rate. Digraph with median as the feature gives good result when compared with other features.