Channel-Wise Spatiotemporal Aggregation Technology for Face Video Forensics

Table 4

Training settings of different parts.

Settings	Training backbones	Training CWSA net
Batch size	32	16
Iteration	50	100
Epoch	40	40
Optimizer	SGD	SGD
Learning rate	0.01	0.01
Momentum	0.9	0.9
Decay	2.375e − 4	2.375e − 4