Instance Space

machine-learning

Definition

Instance Space

The instance space, denoted $\mathcal{X}$, is the set of all possible input objects for a learning algorithm. Each element $x\in \mathcal{X}$, termed an instance or feature vector, represents a specific observation in a $d$-dimensional space, typically ${\mathbb{R}}^d$. The instance space defines the domain of the hypothesis class $\mathcal{H}$ in supervised learning.