Backpropagation

Backpropagation is an algorithm for training neural networks that uses gradient descent to adjust weights in order to minimize error between actual output values and desired outputs.