The core algorithm for training a Neural Network. It computes the Gradient of the Loss Function with respect to the weights of the network.
Backpropogation refers only to the algorithm that computes the gradient. See optimizers for how it is used to change model parameters.
How it works:
It applies the chain rule recursively from the output layer back to the input layer. This results in a vector of partial derivatives that represents how much the Loss Function changes with respect to each parameter.
It does NOT change the parameters itself. That is what optimizers do.