Dropout during pre-training?

I know of the possibility of using the dropout technique in combination with the pre-training.

However, is there anybody who has used it or knows about using dropout during pre-training of an autoencoder?