An adaptive Metropolis algorithm

By Heikki Haario, Eero Saksman, Johanna Tamminen. 2001

2,755 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Mathematical Physics, Statistics and Probability, Probabilistic & Bayesian Methods

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Mathematical Physics

In Statistics and Probability

In Probabilistic & Bayesian Methods

Citations in the corpus

Generalizing Hamiltonian Monte Carlo with Neural Networks (2017, methodology): “ultiple Try Metropolis schemes (Martino & Read, 2013) by having several parametric operators and training each one when used. In addition, one could draw inspiration from the adaptive literature (Haario et al., 2001; Andrieu & Thoms, 2008) or component-wise strategies (Gilks & Wild, 1992”
An Introduction to Probabilistic Programming (2018, methodology): “One way of doing this adaptively, analogous to estimating posterior standard deviations to be used as scale parameters in univariate proposals, is to estimate posterior covariance matrices and using these for jointly proposing multiple latent variables (Haario et al., 2001).”
Kernel Mean Embedding of Distributions: A Review and Beyonds (2016, methodology): “Adaptive MCMC is a class of MCMC methods that try to learn an optimal proposal distribution from previously accepted samples (Haario et al. 2001, Andrieu and Thoms 2008).”
Kernel Adaptive Metropolis-Hastings (2013, background): “conditioned. Intuitively, a good approach to settingγis to slowly decrease it with each adaptation, such that the learned covariance progressively dominates the proposal. Symmetry of the proposal. In Haario et al. (2001), the proposal distribution is asymptotically symmetric due to the vanishing ada”
Coresets for Scalable Bayesian Logistic Regression (2016, methodology): “For the synthetic (resp. real-world) data, adaptive MALA was run once on the full dataset for 100,000 (resp.”
Surrogating the surrogate: accelerating Gaussian-process-based global optimization with a mixture cross-entropy algorithm (2010, methodology): “On the one hand, the optimized surrogate GP covariance could lead to an adaptive initialization of the CE mixture components, similarly to adaptive Metropolis-Hastings with Gaussian proposals (Haario et al., 1998).”
Estimation of ECHAM5 climate model closure parameters with adaptive MCMC (2010, methodology): “These methods included Adaptive Metropolis (AM), the Single Component Adaptive Metropolis (SCAM) and the Delayed Rejection Adaptive Metropolis (DRAM), see (Haario et al., 2001, 2004, 2005, 2006; Andrieu and Moulines, 2006).”
Accelerating Markov Chain Monte Carlo Simulation by Differential Evolution with Self-Adaptive Randomized Subspace Sampling (2009, methodology): “Examples of such methods are the Adaptive Proposal (AP) and Adaptive Metropolis (AM) schemes presented in [15] [16], respectively.”
On adaptive Markov chain Monte Carlo algorithms (2005, methodology): “The set-up is a generalization of Haario et al. (2001).”
An application of Bayesian analysis and Markov chain Monte Carlo methods to the estimation of a regional trend in annual maxima (2006, methodology): “As an illustration, Haario et al. [2001] noted concerning their adaptive AP sampler that ‘‘for many practical applications, the error produced by the AP algorithm is, however, ignorable,’’ although a counterexample can be constructed.”
MCMC methods for multi-response generalized linear mixed models (2010, methodology): “For computational efficiency we use the same M for each block j, where M is the average posterior (co)variance of lj within blocks and is updated each iteration of the burn-in period Haario et al. (2001).”