Backpropagation (Part 0): Rationale - Why Learn Backprop

Backpropagation, the classic stochastic gradient descent algorithm for training a neural network, works hand-in-hand with the Boltzmann machine learning algorithm (contrastive divergence) to enable deep learning. The backpropagation is easier to learn, requiring only repeated application of the chain rule from differential calculus. By mastering backprop, you gain a solid foundation for contrast-and-compare between backpropagation and the Boltzmann machine's contrastive divergence. That gives you a solid understanding of the fundamentals of deep learning systems.

Видео Backpropagation (Part 0): Rationale - Why Learn Backprop канала Alianna J. Maren