DeepLearning5.5, Automatic Differentiation: BackProp revisited

views comments

Optimizing the filters of a convolutional layer by gradient descent is a special case of of automatic differentiation. The basic ideas of automatic differentiation are discussed.