An Empirical Study on GAN-Based Traffic Congestion Attack Analysis: A Visualized Method

Table 5

, , , and of CycleGAN and pix2pix under different learning rates (LR).

Epoch number ()/epoch number (linear decay)	50/50		50/150		100/100		150/50		400/100		250/250		100/400
Framework	CycleGAN	pix2pix	CycleGAN	pix2pix	CycleGAN	pix2pix	CycleGAN	pix2pix	CycleGAN	pix2pix	CycleGAN	pix2pix	CycleGAN	pix2pix
	0.0317	0.1607	0.0114	0.1003	0.0267	0.0929	0.0267	0.1117	0.0113	0.0970	0.0190	0.1070	0.0480	0.1043
	0.0342	0.1674	0.0134	0.1017	0.0340	0.0985	0.0297	0.1118	0.0147	0.1037	0.0250	0.1160	0.0497	0.1126
	2.7000	5.8000	0.5333	3.3400	1.1250	2.4750	0.8000	3.7800	0.5600	6.7200	0.8600	6.5200	1.4400	6.5200
	2.7917	6.0027	0.6245	3.4822	1.5882	2.9441	1.0677	3.8499	0.7072	6.7382	1.0536	6.5437	1.5849	6.5403