Table 2.

MLP baseline training model. Bold indicates the latent layer.

Id	Layer	Type	Output Shape	Activation	Param #
IN	input	Input Layer	(None, 196,608)	ReLU	0
EN256	encoded_256	Dense	(None, 256)	ReLU	50,331,904
EN128	encoded_128	Dense	(None, 128)	ReLU	32,896
L	latent	Dense	(None, 64)	ReLU	8256
DE128	decoded_128	Dense	(None, 128)	ReLU	8320
DE256	decoded_256	Dense	(None, 256)	ReLU	33,024
OUT	output	Dense	(None, 196,608)	Sigmoid	50,528,256