Game controls
Configure a neuron, choose a challenge, then submit your answer.
Keyboard: Enter submit, N new round, H hint, R randomise.
Input features and weights
Edit values or add up to eight inputs.
z = Σ(xᵢwᵢ) + b
Interactive neuron canvas
Connection thickness reflects weight magnitude. Dashed red lines represent negative weights.
Current challenge
Start a new round to receive a challenge.
Your feedback will appear here.
Round progress
0 / 10
Step-by-step calculation
- Choose Step calculation or submit an answer.
Plotly graphs and analytics
Explore activation shape, input contributions, sensitivity, and performance history.
Learning reference
A neuron multiplies each feature by its weight. It adds those contributions and the bias. The activation function transforms that pre-activation value.
z = x₁w₁ + x₂w₂ + ... + xₙwₙ + b and y = f(z)
Positive weights increase the sum when features are positive. Negative weights can reduce it. Bias shifts the activation threshold independently of features.
| Function | Formula | Output range | Typical purpose |
|---|
- Calculate every feature contribution before adding the bias.
- Keep unrounded values until the final answer.
- Watch sigmoid and tanh saturation at large magnitudes.
- Remember that ReLU returns zero for negative inputs.
- Use sensitivity graphs to understand parameter influence.