Randomly sets a fraction rate of input units to 0 at each update during training
time, which helps prevent overfitting.
dropout(rate = 0.5)The fraction of affected units
A construct of class "ruta_network"
Other neural layers:
conv(),
dense(),
input(),
layer_keras(),
output(),
variational_block()