Intrusion detection system combined enhanced random forest with SMOTE algorithm

EURASIP Journal on Advances in Signal Processing

Table 2 Comparison of performance with algorithms on the training set (5-class)

Type	Accuracy (%)				Precision (%)				Recall (%)
Type	KNN	SVM	RF	Our method	KNN	SVM	RF	Our method	KNN	SVM	RF	Our method
Normal	92.47	99.63	98.96	99.91	99.32	98.02	98.41	99.68	99.65	95.62	98.19	99.57
Probe	94.23	99.68	98.74	99.74	97.42	96.77	98.02	96.02	96.19	94.97	84.06	98.54
DOS	99.65	99.57	99.59	99.99	99.76	89.95	96.25	99.88	99.79	94.23	97.87	99.93
U2R	99.96	99.88	99.91	99.92	91.72	76.42	77.54	82.37	76.25	82.22	78.75	91.25
R2L	96.08	99.45	93.37	99.89	93.85	87.26	89.25	96.32	84.49	79.85	77.09	93.10