1 / 12

Introduction to sampling

Introduction to sampling . Discussion on An Introduction to MCMC for Machine Learning, Andrieu et al., 2001. Sampling. What is sampling? Useful for? Bayesian inference and learning Normalization Marginalization Expectation Optimization Model selection. Sampling.

irving
Download Presentation

Introduction to sampling

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Introduction to sampling Discussion on An Introduction to MCMC for Machine Learning, Andrieu et al., 2001

  2. Sampling • What is sampling? • Useful for? • Bayesian inference and learning • Normalization • Marginalization • Expectation • Optimization • Model selection

  3. Sampling • Monte Carlo principle (pg. 5) • Law of large numbers • Central limit theorem

  4. Rejection sampling • Rejection • Drawbacks?

  5. Importance sampling • Importance • Drawbacks?

  6. Importance sampling • {ui, wi }: Sampled representation off(u) • Expectation under f(u)

  7. Markov chains • Homogeneous: • T is time-invariant • Represented using a transition matrix Series of samples such that

  8. Markov chains • Stationary distribution • Conditions for stationary distribution • Irreducible? • Aperiodic? • Detailed balance • Sufficient condition for stationarity of p

  9. MCMC • Markov Chain Monte Carlo • Markov Chain • Monte Carlo • Metropolis Hastings • Special cases • Independent sampler • Metropolis algorithm

  10. Metropolis-Hastings • Target distribution: p(x) • Set up a Markov chain with stationary p(x) • Resulting chain has the desired stationary • Detailed balance Propose (Easy to sample from q) with probability otherwise

  11. Metropolis-Hastings • “Mixing”

  12. Gibbs sampler • Idea • Proposals • Acceptance probability • Always possible?

More Related