An LSTM-Autoencoder Architecture for Anomaly Detection Applied on Compressors Audio Data

Table 12

Model dimensions and operations view for stacked LSTM employing SoftMax layer (created by authors).

Operation		Data	Dimensions	Weights ()	Weights (%)
Input	#####	2	259
LSTM	LLLLL	—	—	62,200	60.5%
tanh	#####	2	50
Dropout	|||	—	—	0	0.0%
	#####	2	50
LSTM	LLLLL	—	—	20,200	19.7%
tanh	#####	2	50
Dropout	|||	—	—	0	0.0%
	#####	2	50
LSTM	LLLLL	—	—	20,200	19.7%
tanh	#####		50
Dropout	|||	—	—	0	0.0%
	#####		50
Dense	XXXXX	—	—	102	0.1%
SoftMax	#####		2