A Survey of Keystroke Dynamics Biometrics

Table 9

Comparison between research works based on numerical inputs.

Study	Data size	Digit length	Method	FAR (%)	FRR (%)	EER (%)	Input device
[24]	7	—	𝑘-nearest neighbor and multilayer perceptron	—	—	78–99*	Infrared virtual num-pad
[25]	9	4	ANOVA	—	—	2.4	Customized num-pad
[82]	20	8	Hidden Markov model	—	—	3.6	Normal keyboard
[137]	5	6	Support vector machine	0.95	5.6	—	Modified keyboard
[48]	10	10	𝑘-nearest neighbor	—	—	1	Notebook touch pad
[26]	30	4	Euclidean distance	15	0	10	Modified ATM num-pad
[158]	10	10	Probabilistic neural network	—	—	1	Notebook touch pad
[33]	28	10	Random forest decision tree	0.03	1.51	1	Apple keyboard
[47]	25	10	Back propagation neural network	—	—	94.8*	Normal keyboard
[4]	24	4	Support vector machine	—	—	2	Modified Microsoft keyboard

 * Indicates performance measurement in terms of accuracy, similar but inverse to EER where value closer to 100% indicates better performance.