Entropy clustering-based granular classifiers for network intrusion detection

EURASIP Journal on Wireless Communications and Networking

Table 6 Comparison results for the 10% KDD Cup 99 dataset

Classifier	Testing samples	TP/FP	DDoS	Probe	U2R	R2L	Mean TP/FP ratio
FNT [27]	11,982	TP	98.75	98.39	99.7	99.09	203.84
FNT [27]	11,982	FP	0.62	1.39	0.22	0.75	203.84
ID3 [28]	311,029	TP	99.52	97.85	49.21	92.75	756.66
ID3 [28]	311,029	FP	0.04	0.55	0.14	10.03	756.66
Naïve Bayes [29]	77,287	TP	79.2	94.8	12.2	0.1	209.01
Naïve Bayes [29]	77,287	FP	1.7	13.3	0.9	0.3	209.01
SVM [30]	15,437	TP	96.32	63.81	34.48	85.64	40.00
SVM [30]	15,437	FP	1.63	0.94	0.05	4.4	40.00
J48 [31]	15,437	TP	96.8	75.2	12.2	0.1	148.75
J48 [31]	15,437	FP	1	0.2	0.1	0.5	148.75
K-means [32]	311,029	TP	97.3	87.6	29.8	6.4	103.86
K-means [32]	311,029	FP	0.4	2.6	0.4	0.1	103.86
SOM [31]	15,437	TP	96.4	74.3	13.3	0.1	125.35
SOM [31]	15,437	FP	0.8	0.3	0.1	0.4	125.35
GAU [32]	311,029	TP	82.4	90.2	22.8	9.6	60.28
GAU [32]	311,029	FP	0.9	11.3	0.5	0.1	60.28
OneR [32]	49,596	TP	94.2	12.9	10.7	10.7	63.80
OneR [32]	49,596	FP	6.8	0.1	2	0.1	63.80
Bayes DT combo [29]	311,029	TP	87.9	76.23	12.33	30.6	44.68
Bayes DT combo [29]	311,029	FP	0.67	1.7	8.9	23.8	44.68
NBTree [31]	15,437	TP	97.4	73.3	1.2	0.1	40.00
NBTree [31]	15,437	FP	1.2	1.1	0.1	0.5	40.00
Naïve Baye s[31]	15,437	TP	79.2	94.8	12.2	0.1	16.90
Naïve Baye s[31]	15,437	FP	1.7	13.3	0.9	0.3	16.90
The proposed ECGC	15,437	TP	99.98	97.85	100	97.57	3055
The proposed ECGC	15,437	FP	0.01	0.05	0	0.89	3055