Skip to main content

Table 5 Accuracy (Acc) rates in percentage and total of selected features (SF) for all the combinations of signature, touch and keyboard

From: An investigation of the predictability of the Brazilian three-modal hand-based behavioural biometric: a feature selection and feature-fusion approach

Feature

Algorithm

Keyboard + touch

Signature + keyboard

Signature + touch

selection

 

Acc (FS)

Acc (FS)

Acc (FS)

No selection

KNN

20.17 (66)

37.71 (74)

43.85 (54)

 

SVM

20.61 (66)

38.15 (74)

42.54 (54)

 

MLP

28.07 (66)

48.68 (74)

42.54 (54)

 

LIBSVM

20.61 (66)

35.52 (74)

35.96 (54)

 

WISARD

51.61 (66)

66.22 (74)

64.91 (54)

Manual

KNN

56.71 (27)

9.64 (11)

10.08 (11)

 

SVM

10.39 (40)

13.59 (5)

32.89 (13)

 

MLP

54.11 (19)

21.05 (26)

16.67 (12)

 

LIBSVM

03.50 (20)

01.31 (37)

01.31 (39)

 

WISARD

18.42 (35)

10.96 (7)

30.26 (17)

CFS

KNN

67.53 (4)

64.47 (10)

62.28 (10)

 

SVM

66.67 (4)

08.77 (10)

9.21 (10)

 

MLP

41.99 (4)

60.08 (10)

58.77 (10)

 

LIBSVM

08.33 (2)

55.26 (10)

53.50 (10)

 

WISARD

60.47 (2)

67.98 (10)

66.22 (10)

RELIEF

KNN

67.53 (5)

60.96 (18)

57.01 (12)

 

SVM

66.67 (5)

60.96 (18)

58.33 (12)

 

MLP

59.31 (5)

60.52 (18)

56.67 (12)

 

LIBSVM

44.29 (33)

57.89 (18)

50.00 (12)

 

WISARD

66.26 (33)

74.12 (18)

68.85 (12)

GA

KNN

74.33 (20)

75.26 (25)

77.63 (19)

 

SVM

48.02 (29)

63.41 (40)

41.83 (27)

 

MLP

33.94 (35)

73.28 (40)

59.60 (32)

 

LIBSVM

8.81 (20)

85.92 (22)

76.05 (8)

 

WISARD

95.91 (32)

96.84 (31)

84.60 (25)