Labeled Data has ground truth annotations (like class labels or regression targets) attached to each sample. It is essential for supervised learning tasks since models learn to map inputs to outputs based on these labels. Acquiring labeled data can be expensive and time-consuming in some domains.