Minsky and Papert showed that single layer perceptrons suffer from exponentially bad scaling to reach a certain accuracy for certain problems. Multi-layer subs

Minsky and Papert showed that single layer perceptrons suffer from exponentially bad scaling to reach a certain accuracy for certain problems. Multi-layer substantially changes the scaling.