Math & Optimization · advanced · concept 77 of 311 · Updated Nov 4, 2026

📈Gaussian Processes

A Gaussian process is a distribution over functions: any finite set of function values is jointly Gaussian, with a kernel deciding how similar nearby inputs should be. Fit to data, it returns not only a prediction at each point but an uncertainty band that widens away from the data. That makes GPs the standard surrogate in Bayesian optimization and a good fit for small, costly datasets. Exact inference is cubic in the number of points.