Machine learning-based network intrusion detection for big and imbalanced data using oversampling, stacking feature embedding and feature extraction

Journal of Big Data

Table 1 Related work summary of various ML techniques

SL. NO.	ML Technique	Algorithm	Author	Dataset	Accuracy (In %)
					Binary	Multi-class
1	ML	RF	[29]	UNSW-NB15	98.67	97.37
2		ELM	[25]			98.43
3		DT	[26]		90.85	67.57
		ANN			84.4	77.51
		LR			77.64	65.29
		KNN			84.46	72.30
		SVM			60.89	53.95
4		C5	[28]			99.3
5		PART	[30]	CIC-IDS2017		99.95
				KDDCUP’99		99.32
6		MapReduce+RF	[31]			93.9
7		PSO+NB	[32]			99.12
8		HFS+LightGBM	[33]	CIC-IDS2018		97.73
9		IG+GR+JRip	[27]	KDDCUP’99		99.57
				BOT-IOT		99.99
10		t-SNERF	[34]	UNSW-NB15		100
				CIC-IDS2017		99.78
				Phishing		99.70
11		DT	[35]	NSL-KDD	99.42
				CIC-IDS2017	98.80
12		IG+RF	[36]		99.86
		IG+J48			99.87
13	Deep Learning	DNN	[37]	UNSW-NB15	99.92	95.9
		RNN			85.42	85.4
		ANN			99.26	97.89
14		DNN	[38]	KDDCUP’99	91.50
				NSL-KDD
				UNSW-NB15
15		LSTM+CNN	[39]	CICIDS-001		99.83
				UNSW-NB15	99.17
16		CNN	[45]	KDDCUP’99		99
				CIC-IDS2018		91.50
17	Hybrid Learning	DNN+ACO	[42]	CIC-IDS2017		98.25
18		CNN+RNN	[43]	CIC-IDS2018		97.75
19		WFEU+FFDN	[44]	AWID	99.66	99.77
				UNSW-NB15	87.10	77.16
20		SGM+CNN	[22]		99.74	96.54
				CIC-IDS2017		99.85
21		WDLSTM+CNN	[45]	UNSW-NB15	97.17	98.43