the visual lab / 01
See the decision boundary.
A small 3D experiment for a big idea: how a perceptron separates two classes.
Loads only when you choose. No animation loop or 3D library.
One plane. Two classes.
For this example, all three weights are 1. The model predicts class 1 when x + y + z ≥ threshold; otherwise it predicts class 0.
Mint: class 1 · Purple: class 0. The translucent plane marks equality. Moving the threshold slides the plane and changes predictions.
This illustrates a fixed rule, not training or a measure of accuracy. A single perceptron can only make a linear separation.
Read the full perceptron lesson →Text explanation
At threshold 0, the point (1, 1, 1) scores 3, so it belongs to class 1. The point (−1, −1, −1) scores −3, so it belongs to class 0. Rotating changes only the camera, never the predictions.