DCAST: A Spatiotemporal Model with DenseNet and GRU Based on Attention Mechanism

Table 2

Parameter settings of the DCAST model.

Parameters	Value
Number of dense blocks	3
Number of GRU layers	2
Number of filters	32
Filters size	(3, 3)
Number of GRU hidden layer neurons	128, 128
Dropout rate	0.5
Learning rate	0.001
Batch size	512
Epoch	150
Optimizer	Adam