Table 6.

Experimental results of the comparative experiment.

Data Category	Recall
	Raw Data (Amount)	Over Sampling (Amount)	Under Sampling (Amount)	CTTGAN (Amount)
BENIGN	0.9864 (100,000)	0.9829 (100,000)	0.9433 (1500)	0.9904 (100,000)
DoS Hulk	1.0000 (10,000)	0.9925 (10,000)	1.0000 (1500)	0.9980 (10,000)
PortScan	0.9990 (10,000)	0.9990 (10,000)	1.0000 (1500)	0.9990 (10,000)
DDoS	0.9975 (10,000)	0.9985 (10,000)	1.0000 (1500)	0.9980 (10,000)
DoS GoldenEye	0.9990 (10,000)	0.9995 (10,000)	0.9967 (1500)	0.9980 (10,000)
FTP-Patator	0.9975 (7894)	0.9981 (7894)	0.9867 (1500)	0.9968 (7894)
SSH-Patator	0.9906 (5861)	0.9915 (5861)	0.9933 (1500)	0.9881 (5861)
DoS slowloris	0.9922 (5771)	0.9931 (5771)	0.9900 (1500)	0.9913 (5771)
DoS slowhttptest	0.9909 (5485)	0.9918 (5485)	0.9967 (1500)	0.9909 (5485)
Bot	0.7918 (1943)	0.9720 (5000)	0.9967 (1500)	0.9980 (5000)
Web Attack Brute Force	0.9431 (1497)	0.9530 (5000)	0.9467 (1497)	0.9960 (5000)

Note: Red indicates the Recall index of small category samples in the CTTGAN scheme, all of which are above 0.99.