Unsupervised Learning
A machine learning approach where the model finds patterns and structure in data without labeled examples, discovering groupings and relationships on its own.
Why it matters
Unsupervised learning powers clustering, anomaly detection, and the embeddings that make semantic search possible. It is how AI finds structure when you do not know what you are looking for.
How it works
Instead of learning from labeled examples, unsupervised learning algorithms look for inherent structure in the data itself. They find clusters of similar items, detect outliers, or compress high-dimensional data into compact representations — all without being told what the "right" answer is.
Common techniques
- Clustering — groups similar data points together. Customer segmentation, document grouping, gene expression analysis.
- Dimensionality reduction — compresses data while preserving important relationships. Visualization, noise removal, feature extraction.
- Anomaly detection — identifies data points that do not fit the expected pattern. Fraud detection, system monitoring, quality control.
Connection to embeddings
The embeddings used in semantic search and RAG systems are a form of unsupervised learning. The model learns to represent text as dense vectors where meaning is encoded in proximity — similar concepts end up near each other in vector space, without anyone labeling which concepts are "similar."