Explore modules

Backpropagation

Visualize the gradient flow and weight updates in a neural network.

Free theory article

Visualize the gradient flow and weight updates in a neural network.

Concept & interpretation

Backpropagation (short for "backward propagation of errors") is the fundamental algorithmic foundation for training artificial neural networks. It computes the gradient of a loss function with respect to the weights of the network, leveraging the chain rule of calculus. These gradients are then utilized by optimization algorithms, such as gradient descent, to systematically adjust the weights and minimize the error.

Read the full free theory article