Unsupervised learning

Unsupervised learning is a machine learning method where models analyze and interpret data without labeled examples. Instead of learning from predefined input-output pairs—or labeled datasets as in supervised learning—the model identifies patterns and relationships in the data on its own.

Unsupervised learning is widely applied in various industries. Common real-world applications include customer segmentation and anomaly detection. It also plays a supporting role in recommendation systems and natural language processing (NLP).

In pre-training, unsupervised learning enables models to extract general-purpose representations from large unlabeled datasets, providing a strong foundation for later task-specific training. After training or deployment, unsupervised methods can also be used to analyze model outputs or embeddings—for example, to cluster results, detect anomalies, or uncover new patterns in unlabeled data.