Perceptron Classifier

The simplest possible learning machine, ported from my Python project. One perceptron, three weights, and a hundred points. Each point is labeled by whether it falls above or below a hidden line, and the perceptron nudges its weights on every point until its guess matches the truth.

• The rule. It sums weighted inputs plus a bias and fires plus one or minus one.
• Learning. Each step it shifts weights toward any point it got wrong.
• Watch it. Green points are classified right, red are wrong, and the purple line is the current boundary.