Keystroke Dynamics User Authentication Based on Gaussian Mixture Model and Deep Belief Nets

Table 1

Performance comparison between the proposed approaches and the existing best reported algorithms on the same CMU dataset. Mean and standard deviation are shown for the equal error rate (EER). The proposed approaches significantly outperform the state-of-the-art approaches.

Algorithm	EER
Neural network (auto-assoc) [11]	0.161 (0.080)
SVM (one-class) [11]	0.102 (0.065)
Manhattan (scaled) [11]	0.096 (0.069)
Combined Mahalanobis and Manhattan distance [12]	0.084 (0.056)
GMM	0.087 (0.058)
GMM-UBM	0.055 (0.052)
DBN	0.035 (0.027)