Create setting for neural network model with python
setMLPTorch( size = c(500, 1000), w_decay = c(5e-04, 0.005), epochs = c(20, 50), seed = 0, class_weight = 0, mlp_type = "MLP", autoencoder = FALSE, vae = FALSE )
size | The number of hidden nodes |
---|---|
w_decay | The l2 regularisation |
epochs | The number of epochs |
seed | A seed for the model |
class_weight | The class weight used for imbalanced data: 0: Inverse ratio between positives and negatives -1: Focal loss |
mlp_type | The type of multiple layer network, inlcuding MLP and SNN (self-normalizing neural network) |
autoencoder | First learn stakced autoencoder for input features, then train MLP on the encoded features. |
vae | First learn stakced varational autoencoder for input features, then train MLP on the encoded features. |
if (FALSE) { model.mlpTorch <- setMLPTorch() }