Variational Inference with Hamiltonian Monte Carlo

Abstract

Variational inference lies at the core of many state-of-the-art algorithms. To improve the approximation of the posterior beyond parametric families, it was proposed to include MCMC steps into the variational lower bound. In this work we explore this idea using steps of the Hamiltonian Monte Carlo (HMC) algorithm, an efficient MCMC method. In particular, we incorporate the acceptance step of the HMC algorithm, guaranteeing asymptotic convergence to the true posterior. Additionally, we introduce some extensions to the HMC algorithm geared towards faster convergence. The theoretical advantages of these modifications are reflected by performance improvements in our experimental results.

References in corpus (6)

• Improving neural networks by preventing co-adaptation of feature detectors
• Stochastic Backpropagation and Approximate Inference in Deep Generative Models
• Variational Inference with Normalizing Flows
• Theano: new features and speed improvements
• DRAW: A Recurrent Neural Network For Image Generation
• General state space Markov chains and MCMC algorithms

Cited by in corpus (8)

• Stochastic Normalizing Flows
• Bayesian Estimation of Mixed Multinomial Logit Models: Advances and Simulation-Based Evaluations
• Learning Symmetries of Classical Integrable Systems
• Variational Bayesian Inference for Mixed Logit Models with Unobserved Inter- and Intra-Individual Heterogeneity
• MetFlow: A New Efficient Method for Bridging the Gap between Markov Chain Monte Carlo and Variational Inference
• MCMC Variational Inference via Uncorrected Hamiltonian Annealing
• Interpolating between sampling and variational inference with infinite stochastic mixtures
• Quasi-symplectic Langevin Variational Autoencoder