Automatic Differentiation Variational Inference (2016-03-02T00:00:00.000000Z)

TL;DR

Automatic differentiation variational inference (ADVI) is developed, where the scientist only provides a probabilistic model and a dataset, nothing else, and the algorithm automatically derives an efficient Variational inference algorithm, freeing the scientist to refine and explore many models.

Abstract

Probabilistic modeling is iterative. A scientist posits a simple model, fits it to her data, refines it according to her analysis, and repeats. However, fitting complex models to large data is a bottleneck in this process. Deriving algorithms for new models can be both mathematically and computationally challenging, which makes it difficult to efficiently cycle through the steps. To this end, we develop automatic differentiation variational inference (ADVI). Using our method, the scientist only provides a probabilistic model and a dataset, nothing else. ADVI automatically derives an efficient variational inference algorithm, freeing the scientist to refine and explore many models. ADVI supports a broad class of models-no conjugacy assumptions are required. We study ADVI across ten different models and apply it to a dataset with millions of observations. ADVI is integrated into Stan, a probabilistic programming system; it is available for immediate use.

Authors

R. Ranganath

3 papers

Dustin Tran

6 papers

D. Blei

8 papers

TL;DR

Abstract

Authors

References86 items

Edward: A library for probabilistic modeling, inference, and criticism

The Generalized Reparameterization Gradient

Theano: A Python framework for fast computation of mathematical expressions

Probabilistic programming in Python using PyMC3

Overdispersed Black-Box Variational Inference

Variational Inference: A Review for Statisticians

Kullback-Leibler Proximal Variational Inference

The Variational Gaussian Process

Hierarchical Variational Models

The Stan Math Library: Reverse-Mode Automatic Differentiation in C++

Fast Second Order Stochastic Backpropagation for Variational Inference

Stochastic gradient variational Bayes for gamma approximating distributions

Weight Uncertainty in Neural Network

Automatic Variational Inference in Stan

Variational Inference with Normalizing Flows

Weight Uncertainty in Neural Networks

Automatic differentiation in machine learning: a survey

Probabilistic Line Searches for Stochastic Optimization

Adam: A Method for Stochastic Optimization

NICE: Non-linear Independent Components Estimation

Doubly Stochastic Variational Bayes for non-Conjugate Inference

Statistical Theory and Inference

Venture: a higher-order probabilistic programming platform with programmable inference

A New Approach to Probabilistic Programming Inference

Stochastic Backpropagation and Approximate Inference in Deep Generative Models

On Using Control Variates with Stochastic Approximation for Variational Bayes and its Connection to Stochastic Linear Regression

Build, Compute, Critique, Repeat: Data Analysis with Latent Variable Models

Black Box Variational Inference

Auto-Encoding Variational Bayes

An Adaptive Learning Rate for Stochastic Variational Inference

Automated Variational Inference in Probabilistic Programming

Affine Independent Variational Inference

Stochastic variational inference

Bayesian reasoning and machine learning

The No-U-turn sampler: adaptively setting path lengths in Hamiltonian Monte Carlo

MCMC Using Hamiltonian Dynamics

Two problems with variational expectation maximisation for time-series models

Riemann manifold Langevin and Hamiltonian Monte Carlo methods

Probability and Stochastics

Adaptive Subgradient Methods for Online Learning and Stochastic Optimization

Bayesian data analysis.

Gaussian Covariance and Scalable Variational Inference

Bayesian Inference for Nonnegative Matrix Factorisation Models

Data Analysis Using Regression and Multilevel/Hierarchical Models

The Variational Gaussian Approximation Revisited

Graphical Models, Exponential Families, and Variational Inference

Church: a language for generative models

Algorithmic Differentiation: Application to Variational Problems in Computer Vision

Validation of Software for Bayesian Models Using Posterior Quantiles

Pattern Recognition and Machine Learning

Variational inference for Dirichlet process mixtures

Variational Message Passing

GaP: a factor model for discrete data

Stochastic volatility: Bayesian computation using automatic differentiation and the extended Kalman filter

Applied Multivariate Statistical Analysis

Learning the parts of objects by non-negative matrix factorization

An Introduction to Variational Methods for Graphical Models

Natural Gradient Works Efficiently in Learning

Stochastic Volatility: Likelihood Inference And Comparison With Arch Models

Unconstrained parametrizations for variance-covariance matrices

Acknowledgments

30 years of adaptive neural networks: perceptron, Madaline, and backpropagation

Conjugate Priors for Exponential Families

Mixtures of Dirichlet Processes with Applications to Bayesian Nonparametric Problems

A Stochastic Approximation Method

Essentials Of Statistical Inference

Stan Modeling Language Users Guide and Reference Manual

Autograd: reverse-mode differentiation of native Python

Stan Development Team

The No-U-turn sampler

Infer.NET 2.6

Overview of the ImageCLEF 2013 Scalable Concept Image Annotation Subtask

Gaussian Kullback-Leibler approximate inference

Variational Bayesian inference for linear and logistic regression

Microsoft Research Cambridge

Stochastic Gradient Descent Tricks