Feature Selection Technique Using Ant Colony Optimization on Keystroke Dynamics

The work is concerned with the use of Ant Colony Optimization algorithm for feature selection of Keystrokes Dynamics and comparison of classification accuracy of Multi-SVM and KNN classifiers. There are various approaches used for feature subset selection but, ACO algorithm gives good performance than other feature selection algorithm like Genetic Based algorithm and Particle Swarm Optimization. In this, first all features are extracted from benchmark dataset, then Multi-SVM and KNN classifiers are trained using all features and their classification accuracy is compared with same training set and test set .Then, features are reduced by Ant Colony Optimization algorithm and then Multi-SVM classifier is trained using reduced features and performance is compared before and after feature selection. The study deals with the use of this technology in ID Password authentication in computer systems, Mail Service Provider and where ID Password is used.