Denoising AutoEncoder

[mau]

One simple question:

if i would simulate classical definition of Denoising AutoEncoder by Vincent, 2008 ( http://machinelearning.org/archive/icml2008/papers/592.pdf ), i can simply put this layers:

flatten -> innerProduct (encode) -> DropOut -> sigmoid -> innerProduct (decode) -> loss

Main point is, differently from classical autoencoder, use DropOut layer to simulate Fill-in-the-blank.
It is right?

Thanks