Table 2 Hyperparameters Configuration
From: Ecosense: a revolution in urban air quality forecasting for smart cities
Hyperparameter | Description | Value |
|---|---|---|
Number of LSTM Layers | Number of stacked LSTM layers in the model | 25 |
Number of Units per Layer | Number of LSTM units per layer | 25 |
Batch Size | Number of samples per gradient update | 64 |
Learning Rate | Step size for weight updates | 0.01 |
Dropout Rate | Fraction of input units to drop during training | 0.03 |
Recurrent Dropout Rate | Fraction of recurrent units to drop during training | 0.05 |
Sequence Length | Number of time steps in each input sequence | 12 |
Optimizer | Algorithm used to optimize the loss function | Adam |
Activation Function | Function used to introduce non-linearity | Softmax |