Control variates for stochastic gradient MCMC
From MaRDI portal
Publication:2329787
Abstract: It is well known that Markov chain Monte Carlo (MCMC) methods scale poorly with dataset size. A popular class of methods for solving this issue is stochastic gradient MCMC. These methods use a noisy estimate of the gradient of the log posterior, which reduces the per iteration computational cost of the algorithm. Despite this, there are a number of results suggesting that stochastic gradient Langevin dynamics (SGLD), probably the most popular of these methods, still has computational cost proportional to the dataset size. We suggest an alternative log posterior gradient estimate for stochastic gradient MCMC, which uses control variates to reduce the variance. We analyse SGLD using this gradient estimate, and show that, under log-concavity assumptions on the target distribution, the computational cost required for a given level of accuracy is independent of the dataset size. Next we show that a different control variate technique, known as zero variance control variates can be applied to SGMCMC algorithms for free. This post-processing step improves the inference of the algorithm by reducing the variance of the MCMC output. Zero variance control variates rely on the gradient of the log posterior; we explore how the variance reduction is affected by replacing this with the noisy gradient estimate calculated by SGMCMC.
Recommendations
- Variance reduction for dependent sequences with applications to stochastic gradient MCMC
- Exploration of the (non-)asymptotic bias and variance of stochastic gradient Langevin dynamics
- Stochastic gradient Markov chain Monte Carlo
- Stochastic gradient Hamiltonian Monte Carlo with variance reduction for Bayesian inference
- Consistency and fluctuations for stochastic gradient Langevin dynamics
Cites work
- 10.1162/jmlr.2003.3.4-5.993
- A Stochastic Approximation Method
- Consistency and fluctuations for stochastic gradient Langevin dynamics
- Exploiting multi-core architectures for reduced-variance estimation with intractable likelihoods
- Exploration of the (non-)asymptotic bias and variance of stochastic gradient Langevin dynamics
- Exponential convergence of Langevin distributions and their discrete approximations
- High-dimensional Bayesian inference via the unadjusted Langevin algorithm
- scientific article; zbMATH DE number 3992765 (Why is no real title available?)
- scientific article; zbMATH DE number 6781368 (Why is no real title available?)
- Large-scale machine learning with stochastic gradient descent
- Robust Stochastic Approximation Approach to Stochastic Programming
- The zig-zag process and super-efficient sampling for Bayesian analysis of big data
- User-friendly guarantees for the Langevin Monte Carlo with inaccurate gradient
- Zero variance differential geometric Markov chain Monte Carlo algorithms
- Zero variance Markov chain Monte Carlo for Bayesian estimators
Cited in
(29)- On sampling from a log-concave density using kinetic Langevin diffusions
- Weak approximation of transformed stochastic gradient MCMC
- Stochastic gradient Hamiltonian Monte Carlo with variance reduction for Bayesian inference
- Nonasymptotic bounds for sampling algorithms without log-concavity
- Consistency and fluctuations for stochastic gradient Langevin dynamics
- Exploration of the (non-)asymptotic bias and variance of stochastic gradient Langevin dynamics
- Variance reduction for dependent sequences with applications to stochastic gradient MCMC
- Stochastic gradient MCMC for state space models
- Parallel Markov chain Monte Carlo for Bayesian hierarchical models with big data, in two stages
- scientific article; zbMATH DE number 7626720 (Why is no real title available?)
- Hamiltonian Monte Carlo with energy conserving subsampling
- Control Variates for the Metropolis–Hastings Algorithm
- Stochastic gradient Markov chain Monte Carlo
- Adaptive step size rules for stochastic optimization in large-scale learning
- Scalable Control Variates for Monte Carlo Methods Via Stochastic Optimization
- Improving sampling accuracy of stochastic gradient MCMC methods via non-uniform subsampling of gradients
- Multi-index antithetic stochastic gradient algorithm
- Efficient and generalizable tuning strategies for stochastic gradient MCMC
- Optimal friction matrix for underdamped Langevin sampling
- Computing Bayes: from then `til now
- SwISS: a scalable Markov chain Monte Carlo divide-and-conquer strategy
- Diffusion approximations and control variates for MCMC
- Regularized zero-variance control variates
- Contraction rate estimates of stochastic gradient kinetic Langevin integrators
- Perturbation analysis of Markov chain Monte Carlo for graphical models
- Stochastic gradient MCMC for nonlinear state space models
- Targeted stochastic gradient MCMC for HMMs with rare latent states
- Unbiased kinetic Langevin Monte Carlo with inexact gradients
- A stochastic conjugate gradient method with restart procedure for machine learning
This page was built for publication: Control variates for stochastic gradient MCMC
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2329787)
