# David M. Blei

professor

Sign in to view more

USA-2015

For contributions to the theory and practice of probabilistic topic modeling and Bayesian machine learning. Press Release

USA-2013

For pioneering the area of topic modeling, which has had profound influence on machine learning foundations as well as industrial practice. David Blei is a leader in the area of large data analysis through machine learning. One of his central contributions is the rigorous treatment of the area of topic modeling, wherein a large data set is mathematically decomposed into latent themes. His approach is based on hierarchical Bayesian models using hidden variables, together with efficient computational methods for estimating model parameters. This work has inspired the work of many others, resulting in new research programs, workshops, and graduate courses. Blei is especially well known for his leadership on the landmark paper (with Jordan and Ng) on Latent Dirichlet Allocation. This work has been cited thousands of times and is used in a range of applications including document summarization, indexing, genomics and image database analysis. It has had definitive impact in a number of web applications at companies ranging from startups to internet giants. Press Release

My research interests include:

Topic models

Probabilistic modeling

Approximate Bayesian inference

## Papers281 papers

The Medical Deconfounder: Assessing Treatment Effect with Electronic Health Records (EHRs).

Deep Generative Models for Highly Structured Data

Using Embeddings to Correct for Unobserved Confounding in Networks.

The Medical Deconfounder - Assessing Treatment Effects with Electronic Health Records.

Avoiding Latent Variable Collapse with Generative Skip Models

Comment: Variational Autoencoders as Empirical Bayes

The Blessings of Multiple Causes: Rejoinder

Proximity Variational Inference.

Augment and Reduce: Stochastic Inference for Large Categorical Distributions.

Variational Sequential Monte Carlo.

Dynamic Embeddings for Language Evolution.

Technical perspective: Expressive probabilistic models and scalable method of moments.

The Blessings of Multiple Causes.

Noisin: Unbiased Regularization for Recurrent Neural Networks.

Empirical Risk Minimization and Stochastic Gradient Descent for Relational Data.

Avoiding Latent Variable Collapse With Generative Skip Models.

A probabilistic approach to discovering dynamic full-brain functional connectivity patterns.

Measuring discursive influence across scholarship.

Readmission prediction via deep contextual embedding of clinical concepts.

A Probabilistic Model of Cardiac Physiology and Electrocardiograms.

Frequentist Consistency of Variational Bayes.

Dynamic Embeddings for Language Evolution.

**1**|Bibtex

The Holdout Randomization Test: Principled and Easy Black Box Feature Selection

The Markov link method: a nonparametric approach to combine observations from multiple experiments

Implicit Causal Models for Genome-wide Association Studies

Automatic Differentiation Variational Inference.

Deep and Hierarchical Implicit Models.

Deep Probabilistic Programming.

Stochastic Gradient Descent as Approximate Bayesian Inference.

Frequentist Consistency of Variational Bayes.

Dynamic Bernoulli Embeddings for Language Evolution.

Zero-Inflated Exponential Family Embeddings.

Robust Probabilistic Modeling with Bayesian Data Reweighting.

Evaluating Bayesian Models with Posterior Dispersion Indices.

Bayesian Learning and Inference in Recurrent Switching Linear Dynamical Systems.

Reparameterization Gradients through Acceptance-Rejection Sampling Algorithms.

Context Selection for Embedding Models.

Structured Embedding Models for Grouped Data.

Hierarchical Implicit Models and Likelihood-Free Variational Inference.

Variational Inference via \chi Upper Bound Minimization.

SHOPPER: A Probabilistic Model of Consumer Choice with Substitutes and Complements.

Comment: A Discussion of “Nonparametric Bayes Modeling of Populations of Networks”

Stochastic Gradient Descent as Approximate Bayesian Inference

Automatic differentiation variational inference

Structured Embedding Models for Grouped Data

Variational Inference via chi Upper Bound Minimization

Modeling User Exposure in Recommendation

Hierarchical Variational Models

Objective Variables for Probabilistic Revenue Maximization in Second-Price Auctions with Reserve

Objective Variables for Probabilistic Revenue Maximization in Second-Price Auctions with Reserve.

Modeling User Exposure in Recommendation.

Reweighted Data for Robust Probabilistic Models.

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

Operator Variational Inference.

Exponential Family Embeddings.

Detecting and Characterizing Events.

The $χ$-Divergence for Approximate Inference.

The Generalized Reparameterization Gradient.

Hierarchical Variational Models.

Operator Variational Inference.

Nested Hierarchical Dirichlet Processes.

A Bayesian Nonparametric Approach to Image Super-Resolution.

Distance Dependent Infinite Latent Feature Models.

Variational inference with copula augmentation.

A Probabilistic Model for Using Social Networks in Personalized Item Recommendation