Neural network gradient Hamiltonian Monte Carlo
From MaRDI portal
Abstract: Hamiltonian Monte Carlo is a widely used algorithm for sampling from posterior distributions of complex Bayesian models. It can efficiently explore high-dimensional parameter spaces guided by simulated Hamiltonian flows. However, the algorithm requires repeated gradient calculations, and these computations become increasingly burdensome as data sets scale. We present a method to substantially reduce the computation burden by using a neural network to approximate the gradient. First, we prove that the proposed method still maintains convergence to the true distribution though the approximated gradient no longer comes from a Hamiltonian system. Second, we conduct experiments on synthetic examples and real data sets to validate the proposed method.
Recommendations
- Manifold stochastic dynamics for Bayesian learning
- Hamiltonian Monte Carlo acceleration using surrogate functions with random bases
- Modified Hamiltonian Monte Carlo for Bayesian inference
- Stochastic gradient Hamiltonian Monte Carlo with variance reduction for Bayesian inference
- Hamiltonian Monte Carlo methods in machine learning
Cites work
- Approximation by superpositions of a sigmoidal function
- Bayesian learning for neural networks
- Emulation of higher-order tensors in manifold Monte Carlo methods for Bayesian inverse problems
- Hamiltonian Monte Carlo acceleration using surrogate functions with random bases
- MCMC using Hamiltonian dynamics
- Simulating Hamiltonian Dynamics
Cited in
(12)- Hamiltonian Monte Carlo acceleration using surrogate functions with random bases
- An adaptive Hessian approximated stochastic gradient MCMC method
- Hamiltonian Markov chain Monte Carlo for partitioned sample spaces with application to Bayesian deep neural nets
- Stochastic gradient Hamiltonian Monte Carlo with variance reduction for Bayesian inference
- Precomputing strategy for Hamiltonian Monte Carlo method based on regularity in parameter space
- Manifold stochastic dynamics for Bayesian learning
- Hamiltonian Monte Carlo methods in machine learning
- Massive Parallelization Boosts Big Bayesian Multidimensional Scaling
- Efficient Bayesian inference with latent Hamiltonian neural networks in no-U-turn sampling
- Hamiltonian Deep Neural Networks Guaranteeing Nonvanishing Gradients by Design
- Efficient Bayes inference in neural networks through adaptive importance sampling
- On the accept-reject mechanism for Metropolis-Hastings algorithms
This page was built for publication: Neural network gradient Hamiltonian Monte Carlo
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1729343)
