def __init__(self, nInputs, nHidden, outputActivation=Sigmoid):
"""
Constructs a new `AutoEncoder` network.
:Parameters:
nInputs : int
The `inputs` size.
nHidden : int
The size of the hidden representation.
outputActivation : class derived from `Activation`
The type of activation for the backprojection layer.
:attention: `outputActivation` parameter is not an instance but a class.
"""
Sequential.__init__(self, nInputs=nInputs)
self.nHidden = nHidden
self.add(Linear(nHidden, nInputs))
self.add(Tanh(nHidden))
self.add(Linear(nInputs, nHidden))
self.add(outputActivation(nInputs))
def __init__(self, nUnits, outputActivation=Sigmoid):
"""
Constructs a new `MultiLayerPerceptron` network.
:Parameters:
nUnits : int list
The sizes of the (input, hidden and output) representations.
outputActivation : class derived from `Activation`
The type of activation for the output layer.
:attention: `outputActivation` parameter is not an instance but a class.
"""
Sequential.__init__(self, nInputs=nUnits[0])
self.nUnits = nUnits
# In and hidden layers
for nOutputs in nUnits[1:-1]:
self.add(Linear(nOutputs))
self.add(Tanh(nOutputs))
# Output layer
self.add(Linear(nUnits[-1]))
self.add(outputActivation(nUnits[-1]))
