Finite Difference vs Backpropagation

This post explores 2 techniques of updating model weights for neural networks - Forward and Backward Propagation. While they both achieve the same end goal of computing derivates of the cost function and updating model weights, there implementation makes a big difference in the runtime and practicality of choosing one over the other in most of the models.