I have a NN that accepts amount spent as input, and predicts the number of clothing items bought as a multi-target regression problem (see: How to count categorization instances in a NN?).
Each clothing item however has a fixed value, and it would make no sense for the NN to predict a sequence of purchased items which has a different value to the input budget. Is there any way I can convey this information to the NN to either force it to confine its output to match the dollar amount of the input, or to assist it in its training?
