A Survey of Keystroke Dynamics Biometrics

Table 6

Comparison between research works in dynamic authentication mode.

Study	Data size	Maximum number of graphs	Freedom of input	Method	FAR (%)	FRR (%)	EER (%)	Platform
[20]	31	di-graph	Yes	Weighted mean, Standard deviation	—	—	90*	OS
[58]	63	di-graph	Yes	𝑘-nearest neighbor	—	—	83.22–92.14*	OS
[59]	205	n-graph	Yes	Degree of disorder, mean, standard deviation	0.5	5	—	Web
[124]	22	di-graph	Yes	Bhattacharyya distance, goodness measure	—	—	86.47*	OS
[74]	61	n-graph	No	SVM and decision tree J48	14.5	1.78	—	OS
[105]	21	di-graph	Yes	Degree of disorder and histogram-based Density estimation	0.14	1.59	—	OS
[72]	10	di-graph	—	Random forest decision tree	0.41	0.63	0.53	—
[107]	21	di-graph	Yes	Random forest decision tree	3.47	0	1.73	Web
[75]	30	di-graph	No	𝑘-nearest neighbor	—	—	0.5	Web
[22]	55	n-graph	Yes	Spearman's foot-rule distance-metric	2.02	1.84	—	Web
[76]	186	di-graph	Yes	Degree of disorder	1.65	2.75	—	Web

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