A GAN and Feature Selection-Based Oversampling Technique for Intrusion Detection

Table 3

Distribution of records in UNSW-NB15 before and after oversampling.

Type	Before oversampling	After oversampling
Normal	37000	37000
Generic	18871	18871
Exploits	11132	11132
Fuzzers	6062	6062
DoS	4089	4089
Reconnaissance	3496	3496
Analysis	677	10677
Backdoor	583	10583
Shellcode	378	10378
Worms	44	10044