Approximate inference methods make it possible to learn realistic models from big data by trading off computation time for accuracy, when exact learning and inference are computationally intractable.

Major methods classes

• Laplace's approximation
• Variational Bayesian methods
• Markov chain Monte Carlo
• Expectation propagation
• Markov random fields
• Bayesian networks
  • Variational message passing
• Loopy and generalized belief propagation

^[1][2]

See also

• Statistical inference
• Fuzzy logic
• Data mining

References

External links

• Tom Minka, Microsoft Research (Nov 2, 2009). "Machine Learning Summer School (MLSS), Cambridge 2009, Approximate Inference" (video lecture).