Backpropagation

Backpropagation computes the gradient of the loss function with respect to the weights of the network. Below is an interactive sandbox visualizing forward activations, backward error gradients, and real-time updates while learning the XOR function.

Sandbox Controls

Learning Rate (η)1.50

Epoch

Total Loss

0.4994

XOR Training Set

Computation Network Graph

Idle / Ready

Positive WeightNegative WeightPositive GradientNegative GradientThickness = Magnitude

Understanding Neural Learning Graph

Our neural network is built with a standard 2 input nodes, 3 hidden neurons, and 1 output neuron layout. Step through the stages to inspect the complete network updates in real-time.

Epoch = dataset cycles. Loss = error size.

Loading Lab...