What Is a Gaussian Process?

Gaussian Process Explainer

A Gaussian process is a probabilistic model that places a distribution over possible functions. Any finite collection of function values follows a joint Gaussian distribution defined by a mean function and a covariance kernel.

After observing training data, the model updates that distribution to produce a predictive mean and predictive uncertainty at new inputs. The kernel expresses assumptions about smoothness, periodicity, scale, and relationships among points.

Gaussian processes are useful for regression, spatial modeling, experimental design, and optimization when uncertainty matters. They can work well with modest datasets and make prior assumptions explicit through their covariance structure.

Exact inference typically requires costly matrix operations that scale poorly with the number of observations. Kernel misspecification, noise assumptions, numerical conditioning, and extrapolation also require care, while sparse approximations can improve scale at some loss of fidelity.

A Gaussian process defines a probability distribution over functions and produces predictions together with uncertainty estimates.

Key terms include Gaussian process, distribution over functions, mean function, covariance kernel, predictive mean, predictive uncertainty and sparse approximation.

Observed data update a prior distribution over functions into a posterior distribution.

Predictions include both a central estimate and uncertainty shaped by data coverage and the chosen kernel.

Exact calculations rely on a kernel matrix whose operations become expensive as observations grow.

Approximations and carefully chosen covariance assumptions make larger applications possible.

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.