Improving the Performance of Machine Learning-Based Network Intrusion Detection Systems on the UNSW-NB15 Dataset

Table 1

A part of UNSW-NB15 dataset distribution.

Class type	Training samples	Training samples percentage	Testing samples	Testing samples percentage
Normal	56000	31.94	37000	44.94
Analysis	2000	1.14	677	0.82
Backdoors	1746	1.00	583	0.71
DoS	12264	6.99	4089	4.97
Exploits	33393	19.05	11132	13.52
Fuzzers	18184	10.37	6062	7.36
Generic	40000	22.81	18871	22.92
Reconnaissance	10491	5.98	3496	4.25
Shellcode	1133	0.65	378	0.46
Worms	130	0.07	44	0.05
Total	175341	100	82332	100