▶Perceptron score
A step activation classifies the sign of a weighted score.
The Role of Weights and Biases
Scaling inputs and shifting thresholds
Weights determine how strongly each feature influences a neuron, while the bias moves the activation threshold independently of the inputs.
watchone idea
→manipulateone example
→leave withone intuition
The essential idea: without bias, every linear decision boundary is forced through the origin.
One Concept · One Example
What to notice
- Weight magnitude sets input importance.
- Weight sign sets directional influence.
- Bias translates the decision boundary.
Rotate with weights, translate with bias
Decision boundary
The zero-score line separates the two output classes.
Why the exam classifier needs a bias
Reproduce the two cases from the slides with fixed weights. Compare the prepared student and the no-preparation student as the bias shifts the pass threshold.
Perceptron from the slidesscore = 1·x₁ + 0.5·x₂ + bPass if score ≥ 0 · Fail otherwise
Prepared studentscore 0.4PASS
No preparationscore -0.7FAIL
Weights rotate and scale a boundary; bias shifts it so the model is not anchored to the origin.