Markov Chain Monte Carlo and Variational Inference: Bridging the Gap

Abstract

Recent advances in stochastic gradient variational inference have made it possible to perform variational Bayesian inference with posterior approximations containing auxiliary random variables. This enables us to explore a new synthesis of variational inference and Monte Carlo methods where we incorporate one or more steps of MCMC into our variational approximation. By doing so we obtain a rich class of inference algorithms bridging the gap between variational methods and MCMC, and offering the best of both worlds: fast posterior approximation through the maximization of an explicit objective, with the option of trading off additional computation for additional accuracy. We describe the theoretical foundations that make this possible and show some promising first results.

References in corpus (7)

• Auto-Encoding Variational Bayes
• Theano: new features and speed improvements
• DRAW: A Recurrent Neural Network For Image Generation
• Neural Variational Inference and Learning in Belief Networks
• Variational Bayesian Inference with Stochastic Search
• A Deep and Tractable Density Estimator
• How to view an MCMC simulation as a permutation, with applications to parallel simulation and improved importance sampling

Cited by in corpus (164)

• Denoising Diffusion Probabilistic Models
• Variational Inference: A Review for Statisticians
• A Review of Uncertainty Quantification in Deep Learning: Techniques, Applications and Challenges
• An Introduction to Variational Autoencoders
• Pixel Recurrent Neural Networks
• Tutorial on Variational Autoencoders
• Variational Inference with Normalizing Flows
• Normalizing Flows: An Introduction and Review of Current Methods
• Weight Normalization: A Simple Reparameterization to Accelerate Training of Deep Neural Networks
• Density estimation using Real NVP
• Adversarially Learned Inference
• KalmanNet: Neural Network Aided Kalman Filtering for Partially Known Dynamics
• Professor Forcing: A New Algorithm for Training Recurrent Networks
• Towards Biologically Plausible Deep Learning
• Multiplicative Normalizing Flows for Variational Bayesian Neural Networks
• Improving Variational Inference with Inverse Autoregressive Flow
• Dynamical Variational Autoencoders: A Comprehensive Review
• Adversarial Variational Bayes: Unifying Variational Autoencoders and Generative Adversarial Networks
• Variational inference for Monte Carlo objectives
• BIVA: A Very Deep Hierarchy of Latent Variables for Generative Modeling
• Data Augmentation in High Dimensional Low Sample Size Setting Using a Geometry-Based Variational Autoencoder
• Time Series Diffusion Method: A Denoising Diffusion Probabilistic Model for Vibration Signal Generation
• Denoising Diffusion Implicit Models
• Hierarchical Implicit Models and Likelihood-Free Variational Inference
• Improved Variational Autoencoders for Text Modeling using Dilated Convolutions
• Seismic tomography using variational inference methods
• Improving Variational Auto-Encoders using Householder Flow
• Inference Suboptimality in Variational Autoencoders
• A-NICE-MC: Adversarial Training for MCMC
• Deep Generative Modelling: A Comparative Review of VAEs, GANs, Normalizing Flows, Energy-Based and Autoregressive Models
• Bias Correction of Learned Generative Models using Likelihood-Free Importance Weighting
• NeuTra-lizing Bad Geometry in Hamiltonian Monte Carlo Using Neural Transport
• On denoising autoencoders trained to minimise binary cross-entropy
• VAE with a VampPrior
• Parameters Estimation for the Cosmic Microwave Background with Bayesian Neural Networks
• $k$-means as a variational EM approximation of Gaussian mixture models
• Variational Sequential Monte Carlo
• The Free Energy Principle for Perception and Action: A Deep Learning Perspective
• Normalizing Flows on Riemannian Manifolds
• Stochastic Normalizing Flows
• Discrete Variational Autoencoders
• Equivariant Hamiltonian Flows
• Simple, Distributed, and Accelerated Probabilistic Programming
• Bayesian Estimation of Mixed Multinomial Logit Models: Advances and Simulation-Based Evaluations
• ZhuSuan: A Library for Bayesian Deep Learning
• Learning Hierarchical Features from Generative Models
• Semi-Amortized Variational Autoencoders
• Machine learning for complete intersection Calabi-Yau manifolds: a methodological study
• A Tutorial on Deep Latent Variable Models of Natural Language
• An introduction to variational inference in Geophysical inverse problems
• Improving Variational Encoder-Decoders in Dialogue Generation
• Learning to Draw Samples with Amortized Stein Variational Gradient Descent
• Diffusion Variational Autoencoders
• On the challenges of learning with inference networks on sparse, high-dimensional data
• Neural Canonical Transformation with Symplectic Flows
• Tighter Variational Bounds are Not Necessarily Better
• Likelihood-free MCMC with Amortized Approximate Ratio Estimators
• Hamiltonian Generative Networks
• Providing physics guides in Bayesian neural networks from input layer: case of giant dipole resonance predictions
• Semi-Implicit Variational Inference
• Advances in Variational Inference
• Approximate Inference with Amortised MCMC
• Generalizing Hamiltonian Monte Carlo with Neural Networks
• Constructing dynamic residential energy lifestyles using Latent Dirichlet Allocation
• Augmented Normalizing Flows: Bridging the Gap Between Generative Flows and Latent Variable Models
• Approximate Bayesian Model Inversion for PDEs with Heterogeneous and State-Dependent Coefficients
• Multiple Causal Inference with Latent Confounding
• Bayesian model reduction
• Fast Second-Order Stochastic Backpropagation for Variational Inference
• Exploiting Hierarchy for Learning and Transfer in KL-regularized RL
• Hierarchical Variational Models
• Designing Perceptual Puzzles by Differentiating Probabilistic Programs
• Learning to superoptimize programs
• Stochastic Normalizing Flows
• Meta-Learning for Stochastic Gradient MCMC
• Particle-based Energetic Variational Inference
• A Review of Learning with Deep Generative Models from Perspective of Graphical Modeling
• Learning Model Reparametrizations: Implicit Variational Inference by Fitting MCMC distributions
• Fast and Accurate Estimation of Non-Nested Binomial Hierarchical Models Using Variational Inference
• Every Smile is Unique: Landmark-Guided Diverse Smile Generation
• Bidirectional Helmholtz Machines
• Reparameterization Gradients through Acceptance-Rejection Sampling Algorithms
• Bayesian Hypernetworks
• PixelVAE++: Improved PixelVAE with Discrete Prior
• Randomized Value Functions via Multiplicative Normalizing Flows
• A likelihood approach to nonparametric estimation of a singular distribution using deep generative models
• Truncated Variational Expectation Maximization
• Markovian Score Climbing: Variational Inference with KL(p||q)
• The Variational Gaussian Process
• Leveraging exploration in off-policy algorithms via normalizing flows
• Adversarial $\alpha$-divergence Minimization for Bayesian Approximate Inference
• Understanding the (un)interpretability of natural image distributions using generative models
• On the Necessity and Effectiveness of Learning the Prior of Variational Auto-Encoder
• Two Methods For Wild Variational Inference
• Bayesian Inference on Binary Spiking Networks Leveraging Nanoscale Device Stochasticity
• Probabilistic Programming with Programmable Variational Inference
• Variational Bayesian Inference for Mixed Logit Models with Unobserved Inter- and Intra-Individual Heterogeneity
• Generate High Resolution Images With Generative Variational Autoencoder
• Deep Markov Chain Monte Carlo
• Pythae: Unifying Generative Autoencoders in Python -- A Benchmarking Use Case
• Deep Variational Inference Without Pixel-Wise Reconstruction
• Geometry-Aware Hamiltonian Variational Auto-Encoder
• Learning Deep Generative Models with Annealed Importance Sampling
• Learning Proposals for Probabilistic Programs with Inference Combinators
• Variational Inference with Continuously-Indexed Normalizing Flows
• MetFlow: A New Efficient Method for Bridging the Gap between Markov Chain Monte Carlo and Variational Inference
• AIDE: An algorithm for measuring the accuracy of probabilistic inference algorithms
• Hierarchical CVAE for Fine-Grained Hate Speech Classification
• Ergodic Inference: Accelerate Convergence by Optimisation
• Hamiltonian Variational Auto-Encoder
• An Uncertain Future: Forecasting from Static Images using Variational Autoencoders
• Annealed Flow Transport Monte Carlo
• Data Augmentation for Enhancing EEG-based Emotion Recognition with Deep Generative Models
• A Neural Network MCMC sampler that maximizes Proposal Entropy
• Accelerating MCMC algorithms through Bayesian Deep Networks
• Sparse Uncertainty Representation in Deep Learning with Inducing Weights
• Variational Inference with Holder Bounds
• CDVAE: Co-embedding Deep Variational Auto Encoder for Conditional Variational Generation
• Variational Rejection Sampling
• Combining Model and Parameter Uncertainty in Bayesian Neural Networks
• Filtering Variational Objectives
• Measuring the non-asymptotic convergence of sequential Monte Carlo samplers using probabilistic programming
• Iterative Refinement of the Approximate Posterior for Directed Belief Networks
• Importance Weighted Hierarchical Variational Inference
• Joint Stochastic Approximation and Its Application to Learning Discrete Latent Variable Models
• Amortized Population Gibbs Samplers with Neural Sufficient Statistics
• Neural Variational Inference and Learning in Undirected Graphical Models
• Energy-Inspired Models: Learning with Sampler-Induced Distributions
• All in the Exponential Family: Bregman Duality in Thermodynamic Variational Inference
• Measure Transport with Kernel Stein Discrepancy
• MCMC Variational Inference via Uncorrected Hamiltonian Annealing
• Interpolating between sampling and variational inference with infinite stochastic mixtures
• Variational Marginal Particle Filters
• General Control Functions for Causal Effect Estimation from Instrumental Variables
• Doubly Semi-Implicit Variational Inference
• Gaussian Mixture Reduction for Time-Constrained Approximate Inference in Hybrid Bayesian Networks
• Channel-Recurrent Autoencoding for Image Modeling
• Quantifying the probable approximation error of probabilistic inference programs
• Gaussian Mean Field Regularizes by Limiting Learned Information
• Soft-Deep Boltzmann Machines
• Generative networks as inverse problems with fractional wavelet scattering networks
• Differentiable Annealed Importance Sampling and the Perils of Gradient Noise
• Variationally Inferred Sampling Through a Refined Bound for Probabilistic Programs
• Quasi-symplectic Langevin Variational Autoencoder
• Cauchy-Schwarz Regularized Autoencoder
• The Theory and Algorithm of Ergodic Inference
• Reparameterized Sampling for Generative Adversarial Networks
• Coarse Grained Exponential Variational Autoencoders
• Contributions to Large Scale Bayesian Inference and Adversarial Machine Learning
• Quantized Variational Inference
• A Hybrid Approximation to the Marginal Likelihood
• Self-Reflective Variational Autoencoder
• Decoupled Variational Embedding for Signed Directed Networks
• Enhanced Variational Inference with Dyadic Transformation
• Learning to Decipher Hate Symbols
• A Divergence Bound for Hybrids of MCMC and Variational Inference and an Application to Langevin Dynamics and SGVI
• Replicating Active Appearance Model by Generator Network
• Improve variational autoEncoder with auxiliary softmax multiclassifier
• Semi-Implicit Stochastic Recurrent Neural Networks
• Transformations between deep neural networks
• MCMC-Interactive Variational Inference
• Structured Stochastic Gradient MCMC
• Probabilistic Autoencoder using Fisher Information
• Hamiltonian Dynamics with Non-Newtonian Momentum for Rapid Sampling