What is a decision boundary in classification?

• A. The hyperplane that separates classes in linear SVMs.
• B. A threshold for converting probabilities to binary labels.
• C. A boundary in feature space that separates different predicted classes.
• D. A decision surface used for clustering tasks.

Answer: (C)

A decision boundary is the surface in the feature space that separates regions where the classifier predicts different classes. It marks where the model changes its mind about the label as you move through the input data, effectively partitioning the space into areas assigned to each class. In a linear SVM, that boundary is a hyperplane, a flat divider, but in other models it can be curved or more complex. It’s not just about converting a score or probability into a label; that thresholding happens in the output space, whereas the decision boundary describes where the actual class predictions switch in the input feature space.