With DropConnect, is inversion calculated on remaining inputs as there is with remaining outputs on DropOut?

Question

In DropOut, the outputs that are not zero\'d, the remaining outputs, have their values divided by probability of keeping the nodes. This scales the values in each layer to offs