What is the term for the distance between the decision boundary and the closest data points from each class?

What is the term for the distance between the decision boundary and the closest data points from each class?

• A. Boundary
• B. Margin
• C. Hyperplane
• D. Threshold

Answer: (B)

The distance from the decision boundary to the closest data points from each class is called the margin. In a linear classifier, the boundary is a hyperplane that separates the classes; the margin measures how thick the gap is between that boundary and the nearest samples on either side. If you express the boundary as w·x + b = 0, the distance from a point x to the boundary is |w·x + b| / ||w||, and the margin is the smallest such distance across all training points (the closest points). Those closest points are the support vectors, and the goal in many approaches, like support vector machines, is to maximize this margin because a larger margin tends to improve generalization by creating a more robust separation. The terms boundary or hyperplane refer to the decision surface itself, not the distance, while a threshold is just a cutoff value used in some classifiers.

The distance from the decision boundary to the closest data points from each class is called the margin. In a linear classifier, the boundary is a hyperplane that separates the classes; the margin measures how thick the gap is between that boundary and the nearest samples on either side. If you express the boundary as w·x + b = 0, the distance from a point x to the boundary is |w·x + b| / ||w||, and the margin is the smallest such distance across all training points (the closest points). Those closest points are the support vectors, and the goal in many approaches, like support vector machines, is to maximize this margin because a larger margin tends to improve generalization by creating a more robust separation. The terms boundary or hyperplane refer to the decision surface itself, not the distance, while a threshold is just a cutoff value used in some classifiers.