Effect of dimensionality reduction on performance in artificial neural network for user authentication

Security is an important concern for today´s generation, where keystroke-scan had come out as a milestone. In this paper, a comparison approach is presented for user authentication using keystroke dynamics. Here we have shown the effect of Dimensionality Reduction techniques on the performance and the misclassification rate is between 9.17% and 9.53%. It helps in improving the performance of the system after reducing the dimensions of input data. We have used three dimensional reduction techniques like: Principal Component Analysis (PCA), Multidimensional scaling (MDS), and probabilistic PCA. Here, PCA provide 9.17% misclassification rate with better performance for keystroke samples of 10 users and each user is having 400 samples for the same password.