Unsupervised learning is a machine learning paradigm where input data is not labelled or pre-classified. Instead, the algorithm must identify patterns and relationships in the data without any prior knowledge or guidance.

This type of learning is commonly used for clustering or grouping data into similar categories or for dimension reduction, where a large dataset is condensed into a smaller set of meaningful features. Unsupervised learning is often used to help identify hidden patterns or insights in a dataset.