Network intrusion detection using feature fusion with deep learning

Journal of Big Data

Table 4 The testing performance of the state-of-the-art machine learning models on the multi-classification task of the UNSW-NB15 and NSL-KDD Dataset

Model	Refs.	Dataset	Accuracy	Precision	Recall
Model	Refs.		(%)	(%)	(%)
Logistic Regression	[18]	UNSW-NB15	65.53	76.91	65.54
Support Vector Machine	[18]	UNSW-NB15	61.09	47.47	62.00
Decision Tree	[18]	UNSW-NB15	66.03	79.82	66.04
scale-hybrid-IDS-AlertNet	[33]	UNSW-NB15	66.00	62.30	66.00
MDNN	[34]	UNSW-NB15	62.87	76.00	63.00
MCNN	[35]	UNSW-NB15	69.46	84.00	69.00
Naive Bayes	[35]	UNSW-NB15	45.22	29.67	38.62
J48	[35]	UNSW-NB15	51.50	28.18	21.48
Random Forest	[35]	UNSW-NB15	68.09	62.51	35.15
Bagging	[35]	UNSW-NB15	51.45	32.85	21.45
Adaboost	[35]	UNSW-NB15	51.50	28.18	21.48
DT & GA	[39]	UNSW-NB15	84.33	53.20	52.23
Our proposed model	–	UNSW-NB15	77.84	86.04	69.50
scale-hybrid-IDS-AlertNet	[33]	NSL-KDD	78.50	81.00	78.50
MDNN	[34]	NSL-KDD	77.55	81.23	77.55
MCNN	[35]	NSL-KDD	81.1	83	81
Naive Bayes	[35]	NSL-KDD	72.73	76.1	72.7
J48	[35]	NSL-KDD	74.99	79.6	75.0
Random Forest	[35]	NSL-KDD	76.45	82.1	76.4
Bagging	[35]	NSL-KDD	74.83	78.3	74.8
AE	[26]	NSL-KDD	81.21	87.85	82.04
LSTM	[26]	NSL-KDD	67.17	82.34	72.49
MLP	[26]	NSL-KDD	68.26	85.05	73.96
L-SVM	[26]	NSL-KDD	69.73	86.77	74.09
Q-SVM	[26]	NSL-KDD	75.11	87.22	77.94
LDA	[26]	NSL-KDD	76.49	80.82	76.72
QDA	[26]	NSL-KDD	64.36	78.26	70.38
Our proposed model	–	NSL-KDD	86.81	86.86	86.80