What Is t-SNE?

t-SNE Explainer

t-distributed stochastic neighbor embedding, or t-SNE, is a visualization method that places high-dimensional observations in two or three dimensions. It tries to keep points that were close in the original representation near one another.

The method converts neighborhood relationships into probability distributions and optimizes the low-dimensional arrangement to make those relationships similar. Random initialization and parameters such as perplexity influence the resulting map.

t-SNE is valuable for exploring embeddings, cell measurements, image features, and other complex representations. It may reveal local groupings that deserve investigation, but the picture is not itself a clustering proof.

Distances between far-apart groups, group sizes, and empty spaces can be misleading. Analysts should test multiple settings and seeds, compare with other methods, retain labels separately, and return to the original data before drawing conclusions.

t-SNE maps high-dimensional observations into two or three dimensions while emphasizing nearby relationships for visualization.

Key terms include t-distributed stochastic neighbor embedding, dimensionality reduction, neighborhood, perplexity, embedding, initialization and visualization.

t-SNE emphasizes which observations are nearby rather than preserving every global distance.

Optimization places similar local neighborhoods together in a map that people can inspect.

Cluster spacing and size can change with settings and should not be read as exact quantities.

Repeated runs and checks against original features are necessary before interpreting apparent groups.

Explore more "Explainers"

Discover additional explainers across politics, science, business, technology, and other fields. Each explainer breaks down a complex idea into clear, everyday language—helping you better understand how major concepts, systems, and debates shape the world around us.