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

Table 6

Model parameters for Stacked GRU architecture (created by authors).

Model Hyperparameters and configuration
Model architecture	Stacked GRU
Number of classes	2
Number of layers	6
Batch size	128
Number of epochs	50
Number of units	50
Dropout rate	0.1
Total number of trainable parameters	77,352