A playground of Boolean operations learned by a single neuron.
A neuron like this is a building block of artificial neural networks. In this simple form, it takes binary inputs (0 or 1), and produces binary output with a formula
\[h(\vec{x})=S(w_0x_0+w_1x_1+w_2x_2)\] \[S(t)=\frac{1}{1+e^{-t}}\]Where S is a sigmoid function:
x₀ is always 1, and is used to apply “bias” weight w₀.
This simple neuron can learn basic Boolean operations:
 |
AND |
 |
OR |
 |
NOT (on first input) |
Use the sliders to change weights, and train the neuron to these basic operations.