Gibbs sampling
Muestreo de Gibbs
- Date
- 1984 · confidence high
- Place
- Providence, Rhode Island, United States
- Credited to
- Stuart Geman y Donald Geman
- Field
- Knowledge & symbols
What it is
Markov chain Monte Carlo (MCMC) algorithm that generates samples from a complex multivariate probability distribution by successively sampling from its conditional distributions, when direct sampling from the joint distribution is impractical.
What it led to
- Standard tool of computational Bayesian inference
- Widespread application in statistics, machine learning and image processing
- Randomized alternative to deterministic algorithms such as expectation-maximization (EM)