Label Space

machine-learning

Definition

Label Space

The label space, denoted $\mathcal{Y}$, is the set of all possible output values (labels) in a supervised learning task. The nature of this space defines the fundamental task of the learner:

• Classification: $\mathcal{Y}$ is a discrete set of categories or nominal labels.
• Regression: $\mathcal{Y}$ is typically a continuous subset of the real numbers $\mathbb{R}$.