A Survey of Keystroke Dynamics Biometrics

Table 5

Comparison between research works with static authentication mode (long inputs).

Study	Data size	Latency	Input repetition	Input freedom	Method	FAR (%)	FRR (%)	EER (%)	Input sessions	Device freedom
[12]	7	FT	1	No	t-Test	—	—	95^†	Yes	No
[40]	43	DT, FT	9	No	Parallel decision trees, Monte Carlo	0.88	9.62	—	Yes	—
[156]	31	FT	2	Yes	Degree of Disorder	1.99	2.42	—	—	No
[87]	—	DT, FT	—	No	Decision tree c4.5, j48	—	—	93.3*	—	—
[77]	118	DT, FT	5	Yes	Euclidean distance	—	—	97.9*	—	Both
[135]	24	DT, FT	60	Yes	Support vector machine	0.76	0.81	1.57	—	—
[88]	112	DT, FT	—	No	Weighted Euclidean distance	—	—	100*	Yes	—
[157]	35	FT	—	No	Kolmogorov-Smirnov test	—	—	7.55	—	No
[34]	51	FT	5	No	Euclidean distance	—	—	0.84	No	—
[115]	33	DT, FT	9	No	Naive Bayesian	—	—	1.72	Yes	Yes
[65]	189	DT, FT	—	No	Weighted Euclidean distance, array disorder	0.01	3	—	No	—
[121]	20	FT	5	Yes	Euclidean distance	2	4	—	No	—
[126]	50	FT	—	Yes	Degree of Disorder	—	—	10	Yes	Yes

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