14-backpropagation

Question 1

Why can’t we use the perceptron rule for neural networks?

Answer 1

The perceptron rule requires the true target outputs, but we only have access to true values at the output layer. As a result we use backpropagation

Question 2

What is a backpropagation?

Answer 2

Backpropagation provides us with a way to update weights in a MLP. It works by propagating the error from the output node through the hidden layer nodes