Unsupervised learning finds patterns or structure in data without being given predefined labels.
Unsupervised learning is a machine-learning method that learns from unlabeled data.
Instead of receiving a correct answer for each example, the system looks for structure such as similar groups, patterns, relationships, and unusual data points.
A clustering algorithm may group similar examples together, but those groups do not automatically come with human names such as “cat” or “dog.”
Sorting a mixed box without labels. Imagine receiving a large box of mixed objects with no labels. You may naturally group together round objects, metal objects, or small objects. No one gives you the group names in advance. The structure comes from similarities in the data.
Customer behavior
↓
Find similar usage patterns
↓
Create customer groups
Many documents
↓
Compare patterns
↓
Group similar topics
Normal activity patterns
↓
Find unusual examples
↓
Flag possible anomalies
Unsupervised: No predefined labels.
Supervised: Training examples include known labels.
The system uses measurable patterns or similarity.
A cluster: A mathematical grouping. A human concept: The meaning people assign to the group.
Further connections: Clustering · Embeddings · Dimensionality reduction.
Explore next
Unsupervised learning finds structure in data without being given the correct labels in advance.