Backpropagation works out how each weight affected an error, so the network can improve.
Why do we need this? A network needs a way to adjust many weights using one measured error.
Positive weightNegative weightActive signal →← Backward gradient
STEP 1 OF 10
Inputs enter the network
The input layer holds the measured features. It passes these values to the hidden layer.
No account required. Local images and model files stay in your browser. Educational models demonstrate mechanisms; they are not production predictors.