in convex optimisation, often machine learning, gradients (partial derivatives) are used. In deep learning an approximation of the chain rule is used in backpro

in convex optimisation, often machine learning, gradients (partial derivatives) are used. In deep learning an approximation of the chain rule is used in backpropigation.