1. A computational method for calculating derivatives of a function by traversing the computational graph in reverse order, from outputs to inputs, commonly used in machine learning for backpropagation
Reverse-mode automatic differentiation is the backbone of modern deep learning frameworks like PyTorch and TensorFlow.
A diferenciação automática em modo reverso é a base dos frameworks modernos de aprendizado profundo, como PyTorch e TensorFlow.
2. A technique that efficiently computes gradients by applying the chain rule backwards through a computational graph
The algorithm uses reverse-mode automatic differentiation to optimize neural network weights.
O algoritmo utiliza diferenciação automática em modo reverso para otimizar os pesos da rede neural.