111
User Authentication
recognition to text classification. Due to the remarkable success rate, SVMs are also used in
keystroke dynamics not only to identify the user but also to recognize the soft biometric
information. A support vector–based machine distinguishes imposter pattern by creating
margin which separates other patterns from that of the imposter, which provides a learning
technique for pattern recognition and regression estimation. It is commonly used and effective for large practical problems. To predict the gender, Giot et al. [7] used libSVM—a library
of SVM. But in our study, we have used FRNN.
As a recognition method, FRNN classification algorithm with vaguely quantified rough
sets is more suitable. This method is an alternative to Sarkar’s fuzzy rough ownership
function (FRNN-O) approach [31]. FRNN uses the nearest neighbors to construct lower
and upper approximations of decision classes, and classifies test instances based on their
membership to these approximations [22]. FRNN-VQRS is a new approach to FRNN.
The hybridization of rough sets and fuzzy sets has focused on creating an end product
that extends both contributing computing paradigms in a conventional way.
TABLE 6.3
Accuracy to Predict the Gender
Classification
Algorithms
Accuracy (%)
Dataset A
Dataset B
Dataset C
Dataset D
FRNN-VQRS [22, 23]
94.81
88.55
95.04
84.75
FRNN [22]
94.81
88.55
95.04
84.75
Fuzzy Rough NN [22]
93.16
85.93
93.45
84.75
Random Forest [24]
92.75
87.54
93.11
79.1
Bagging [25]
91.34
85.59
91.42
76.97
Fuzzy NN [22]
88.92
81.85
92.64
76.45
IBK (Euclidean) [26]
88.71
80.83
91.72
81.07
J48 [27]
86.33
80.34
88.28
71.82
MLP [28]
82.15
75.71
85.89
68.24
SVM [29]
71.47
69.38
79.16
65.72
Naive Bayes [30]
64.37
63.9
72.11
56.7
TABLE 6.4
Accuracy to Predict the Age Group
Classification
Algorithms
Accuracy (%)
Dataset A
Dataset B
Dataset C
Dataset D
FRNN-VQRS [22, 23]
94.31
86.87
94.68
84.75
FRNN [22]
94.31
86.87
94.68
84.75
Fuzzy Rough NN [22]
92.81
83.10
93.24
79.70
Random Forest [24]
92.13
86.40
92.47
79.81
Bagging [25]
90.65
84.63
90.09
75.60
Fuzzy NN [22]
88.03
78.62
92.04
72.45
IBK (Euclidean) [26]
88.00
76.79
91.22
77.92
J48 [27]
86.35
77.89
86.53
64.88
MLP [28]
79.74
70.39
84.55
73.08
SVM [29]
65.41
58.71
70.59
66.67
Naive Bayes [30]
59.13
56.18
66.99
59.41
User Authentication
recognition to text classification. Due to the remarkable success rate, SVMs are also used in
keystroke dynamics not only to identify the user but also to recognize the soft biometric
information. A support vector–based machine distinguishes imposter pattern by creating
margin which separates other patterns from that of the imposter, which provides a learning
technique for pattern recognition and regression estimation. It is commonly used and effective for large practical problems. To predict the gender, Giot et al. [7] used libSVM—a library
of SVM. But in our study, we have used FRNN.
As a recognition method, FRNN classification algorithm with vaguely quantified rough
sets is more suitable. This method is an alternative to Sarkar’s fuzzy rough ownership
function (FRNN-O) approach [31]. FRNN uses the nearest neighbors to construct lower
and upper approximations of decision classes, and classifies test instances based on their
membership to these approximations [22]. FRNN-VQRS is a new approach to FRNN.
The hybridization of rough sets and fuzzy sets has focused on creating an end product
that extends both contributing computing paradigms in a conventional way.
TABLE 6.3
Accuracy to Predict the Gender
Classification
Algorithms
Accuracy (%)
Dataset A
Dataset B
Dataset C
Dataset D
FRNN-VQRS [22, 23]
94.81
88.55
95.04
84.75
FRNN [22]
94.81
88.55
95.04
84.75
Fuzzy Rough NN [22]
93.16
85.93
93.45
84.75
Random Forest [24]
92.75
87.54
93.11
79.1
Bagging [25]
91.34
85.59
91.42
76.97
Fuzzy NN [22]
88.92
81.85
92.64
76.45
IBK (Euclidean) [26]
88.71
80.83
91.72
81.07
J48 [27]
86.33
80.34
88.28
71.82
MLP [28]
82.15
75.71
85.89
68.24
SVM [29]
71.47
69.38
79.16
65.72
Naive Bayes [30]
64.37
63.9
72.11
56.7
TABLE 6.4
Accuracy to Predict the Age Group
Classification
Algorithms
Accuracy (%)
Dataset A
Dataset B
Dataset C
Dataset D
FRNN-VQRS [22, 23]
94.31
86.87
94.68
84.75
FRNN [22]
94.31
86.87
94.68
84.75
Fuzzy Rough NN [22]
92.81
83.10
93.24
79.70
Random Forest [24]
92.13
86.40
92.47
79.81
Bagging [25]
90.65
84.63
90.09
75.60
Fuzzy NN [22]
88.03
78.62
92.04
72.45
IBK (Euclidean) [26]
88.00
76.79
91.22
77.92
J48 [27]
86.35
77.89
86.53
64.88
MLP [28]
79.74
70.39
84.55
73.08
SVM [29]
65.41
58.71
70.59
66.67
Naive Bayes [30]
59.13
56.18
66.99
59.41