How is supervised learning different from unsupervised learning?

Supervised learning is fundamentally characterized by its use of labeled data, which means that each training example is paired with a corresponding output label. This allows the algorithm to learn from the input-output relationships and make predictions based on new, unseen data. An example would be a classification task where an algorithm learns to identify whether images contain cats or dogs based on labeled examples.

In contrast, unsupervised learning involves training algorithms on data that does not have any labels, meaning the algorithm seeks to find patterns or structures within the data without explicit guidance on what to look for. For instance, clustering algorithms might group similar customer behaviors together without knowing in advance what those behaviors represent.

This distinction in how the data is treated defines the core difference between the two learning paradigms. The other options explore notions that are not universally true or do not accurately encapsulate the primary distinction between supervised and unsupervised learning.