Dataset

Preset Click mode

Training

Learning rate η: 0.10 Speed: 8 steps/s Max iterations: 2000

Stats

Positive (+1)0

Negative (−1)0

Iteration0

Mistakes (pass)0

Total mistakes0

w₁0.000

w₂0.000

bias b0.000

Margin—

Convergedno

Rosenblatt's rule: for a misclassified sample (x, y), update w ← w + η·y·x, b ← b + η·y. Predict sign(w·x + b). On linearly separable data the perceptron provably converges; on XOR it never settles.