A decision boundary is the surface in the feature space that separates different classes according to a classifier’s learned decision rule. For a simple model like a linear classifier in 2D, it’s a line (e.g., w·x + b = 0). More complex models yield nonlinear boundaries (curves, complex surfaces). The decision boundary divides the space into regions where the model predicts each possible class. For instance, in a binary classification, one side of the boundary is class A, the other is class B. The margin in SVMs, for example, is related to the distance of training points from the decision boundary. Visualizing decision boundaries (when possible) helps in understanding model behavior – points near the boundary are less confidently classified than those far away.