Home▸AI & Machine Learning▸Neural Network Backpropagation Simulator (2D)

Neural Network Backpropagation Simulator (2D)

A real 2-4-1 feed-forward network learns the XOR problem on a flat 2D node-link diagram: live activations, weight-sign coloring, and loss/accuracy you control.

AI & Machine Learning2DModerate60 FPS📱 Mobile-adapted⇄ 3D version

This 2D companion runs the exact same backpropagation math as the 3D version — a 2-4-1 network training on XOR by batch gradient descent — but draws it as a flat node-link diagram instead of an orbiting 3D scene, so weight signs and neuron activations read at a glance while you tune learning rate, training speed and hidden-layer size.

⚙ Under the hood

Real feed-forward network (2→N→1) trained live with sigmoid activations and gradient descent on the XOR dataset; weight color/thickness encode sign and magnitude, a pulse traces the forward pass.

neural networkbackpropagationgradient descentxor problemmachine learningsigmoid

2D · HTML5 Canvas 2D · 60 FPS target · runs fully client-side, no install