Jacobian matrix
Jacobian matrix: The matrix of first derivatives of every output of a vector-valued function with respect to every input. It generalises gradients to multiple outputs and appears in the chain rule. Automatic differentiation often avoids materialising it fully. This perspective connects the definition with practical behaviour in artificial intelligence systems.
In italiano: Matrice JacobianaLegge IV · Loss, gradiente e ottimizzatoriIt generalises gradients to multiple outputs and appears in the chain rule. Automatic differentiation often avoids materialising it fully.