Neural Network Explorer

Single Neuron — Activation Surface

Adjust weights and bias to see how a single sigmoid neuron draws a linear decision boundary through the XOR input space.

Parameters
Weight A  (wa) 4.0
Weight B  (wb) 4.0
Bias  (b) −2.0
Forward pass
ŷ = ...
P̂ = σ(ŷ)
(a=0, b=0) · true: 0
(a=0, b=1) · true: 1
(a=1, b=0) · true: 1
(a=1, b=1) · true: 0
Why can't one neuron solve XOR?
A single neuron draws one straight line. XOR requires two — no single linear boundary separates the classes correctly. Red border = misclassified point.
Activation surface & decision boundary
label = 0 (circle)
label = 1 (square)
0.5 boundary
misclassified