Neural Network Playground

What is a neural network? At its heart, it's a mathematical function inspired by the brain, designed to recognize patterns. It's built from layers of simple units called 'neurons'. This playground lets you build your own network from the ground up and see exactly how it processes information, one layer at a time.

Section 1: The Input Layer

The network's journey begins at the Input Layer. Each neuron here represents a single piece of input data. A neuron's value, called its 'activation', is just a number between 0 (off/black) and 1 (on/white). Use the sliders below to set the initial activations for your network.

Number of Input Neurons: 3

Section 2: Hidden & Output Layers

The real work happens in the Hidden Layers. These intermediate layers allow the network to combine the initial inputs in increasingly complex ways. The final Output Layer gives the result. The neuron with the highest activation in this last layer is the network's 'guess'.

Number of Hidden Layers: 1

Neurons per Hidden Layer: 4

Number of Output Neurons: 2

Section 3: How Neurons "Think"

A neuron calculates a weighted sum of all incoming activations. It adds a personal 'bias', then passes the result through an Activation Function to 'squish' it into a useful range (e.g., 0 to 1).

Activation = σ(Weighted Sum + Bias)

Activation Function:

Activation Function Graph:

Hover over any connection line to see its weight. Green is positive, red is negative. Thicker lines have a stronger influence.

Hover over any hidden/output neuron to see its bias and the calculation that determines its activation.

Click a connection or neuron to manually edit its weight or bias. Try setting a weight to a large number like 10 and see how it dominates the calculation!