Table 12 Structure of Neural Network
Input | Ulsan real-time dataset |
|---|---|
Learning | Optimizer = ‘adam’, loss = ‘mean_squared_error’ Learning rate = 0.07 Batch size = 150 Verbose = 1 Validation split = .2 |
Hyper Parameters | Scaler = Min_Max Scaler (feature_range = (0, 1)), Standard Scaler Train-test split ratio = Training size (0.8), Test size (0.2) Cov_mat = (14, 14) Number of iteration = 30 epochs Matrix = ‘MAE’, ‘MSE’ Layers = Dense, LSTM Units = 50 Loss = Mean Squared Error |
Output | Predicted result image (14, 50) shape |