Resampling imbalanced data for network intrusion detection datasets

Journal of Big Data

Table 19 Resampling of UNSW-NB15

UNSW-NB15		Number of cases for training model
Category	Label	Before resampling	RU	RO	RURO
Benign	0	1,552,663	31,313	1,552,663	31,313
Fuzzers	1	3563	3563	31,313	31,313
Reconnaissance	2	1217	1217	31,313	31,313
Shellcode	3	165	165	31,313	31,313
Analysis	4	1790	1790	31,313	31,313
Backdoors	5	361	361	31,313	31,313
DoS	6	11,439	11,439	31,313	31,313
Exploits	7	31,313	31,313	31,313	31,313
Generic	8	151,074	31,313	151,074	31,313
Worms	9	130	130	31,313	31,313