Timezone: »
Modern variational inference (VI) uses stochastic gradients to avoid intractable expectations, enabling large-scale probabilistic inference in complex models. VI posits a family of approximating distributions q and then finds the member of that family that is closest to the exact posterior p. Traditionally, VI algorithms minimize the “exclusive Kullback-Leibler (KL)” KL(q||p), often for computational convenience. Recent research, however, has also focused on the “inclusive KL” KL(p||q), which has good statistical properties that makes it more appropriate for certain inference problems. This paper develops a simple algorithm for reliably minimizing the inclusive KL using stochastic gradients with vanishing bias. This method, which we call Markovian score climbing (MSC), converges to a local optimum of the inclusive KL. It does not suffer from the systematic errors inherent in existing methods, such as Reweighted Wake-Sleep and Neural Adaptive Sequential Monte Carlo, which lead to bias in their final estimates. We illustrate convergence on a toy model and demonstrate the utility of MSC on Bayesian probit regression for classification as well as a stochastic volatility model for financial data.
Author Information
Christian Naesseth (Columbia University)
Fredrik Lindsten (Linköping University)
David Blei (Columbia University)
David Blei is a Professor of Statistics and Computer Science at Columbia University, and a member of the Columbia Data Science Institute. His research is in statistical machine learning, involving probabilistic topic models, Bayesian nonparametric methods, and approximate posterior inference algorithms for massive data. He works on a variety of applications, including text, images, music, social networks, user behavior, and scientific data. David has received several awards for his research, including a Sloan Fellowship (2010), Office of Naval Research Young Investigator Award (2011), Presidential Early Career Award for Scientists and Engineers (2011), Blavatnik Faculty Award (2013), and ACM-Infosys Foundation Award (2013). He is a fellow of the ACM.
More from the Same Authors
-
2021 : Unveiling Mode-connectivity of the ELBO Landscape »
Edith Zhang · David Blei -
2022 : An Invariant Learning Characterization of Controlled Text Generation »
Claudia Shi · Carolina Zheng · Keyon Vafa · Amir Feder · David Blei -
2022 : A Bayesian Causal Inference Approach for Assessing Fairness in Clinical Decision-Making »
Linying Zhang · Lauren Richter · Yixin Wang · Anna Ostropolets · Noemie Elhadad · David Blei · George Hripcsak -
2022 : Adjusting the Gender Wage Gap with a Low-Dimensional Representation of Job History »
Keyon Vafa · Susan Athey · David Blei -
2022 : A Modelling Framework for Catalysing Progress in the Rod-Shaped Bacterial Cell Growth Discourse »
Shashi Nagarajan · Fredrik Lindsten -
2022 : CAREER: Economic Prediction of Labor Sequence Data Under Distribution Shift »
Keyon Vafa · Emil Palikot · Tianyu Du · Ayush Kanodia · Susan Athey · David Blei -
2022 : An Invariant Learning Characterization of Controlled Text Generation »
Claudia Shi · Carolina Zheng · Keyon Vafa · Amir Feder · David Blei -
2022 : CAREER: Economic Prediction of Labor Sequence Data Under Distribution Shift »
Keyon Vafa · Emil Palikot · Tianyu Du · Ayush Kanodia · Susan Athey · David Blei -
2022 : An Invariant Learning Characterization of Controlled Text Generation »
Claudia Shi · Carolina Zheng · Keyon Vafa · Amir Feder · David Blei -
2021 : David Blei - On the Assumptions of Synthetic Control Methods »
David Blei -
2021 Test Of Time: Online Learning for Latent Dirichlet Allocation »
Matthew Hoffman · Francis Bach · David Blei -
2021 Poster: Posterior Collapse and Latent Variable Non-identifiability »
Yixin Wang · David Blei · John Cunningham -
2020 Workshop: I Can’t Believe It’s Not Better! Bridging the gap between theory and empiricism in probabilistic machine learning »
Jessica Forde · Francisco Ruiz · Melanie Fernandez Pradier · Aaron Schein · Finale Doshi-Velez · Isabel Valera · David Blei · Hanna Wallach -
2019 Poster: Calibration tests in multi-class classification: A unifying framework »
David Widmann · Fredrik Lindsten · Dave Zachariah -
2019 Spotlight: Calibration tests in multi-class classification: A unifying framework »
David Widmann · Fredrik Lindsten · Dave Zachariah -
2019 Poster: Parameter elimination in particle Gibbs sampling »
Anna Wigren · Riccardo Sven Risuleo · Lawrence Murray · Fredrik Lindsten -
2019 Poster: Poisson-Randomized Gamma Dynamical Systems »
Aaron Schein · Scott Linderman · Mingyuan Zhou · David Blei · Hanna Wallach -
2019 Poster: Pseudo-Extended Markov chain Monte Carlo »
Christopher Nemeth · Fredrik Lindsten · Maurizio Filippone · James Hensman -
2019 Poster: Variational Bayes under Model Misspecification »
Yixin Wang · David Blei -
2019 Poster: Using Embeddings to Correct for Unobserved Confounding in Networks »
Victor Veitch · Yixin Wang · David Blei -
2019 Oral: Parameter elimination in particle Gibbs sampling »
Anna Wigren · Riccardo Sven Risuleo · Lawrence Murray · Fredrik Lindsten -
2019 Poster: Adapting Neural Networks for the Estimation of Treatment Effects »
Claudia Shi · David Blei · Victor Veitch -
2018 : Datasets and Benchmarks for Causal Learning »
Csaba Szepesvari · Isabelle Guyon · Nicolai Meinshausen · David Blei · Elias Bareinboim · Bernhard Schölkopf · Pietro Perona -
2018 : The Blessings of Multiple Causes »
David Blei -
2017 : Panel: On the Foundations and Future of Approximate Inference »
David Blei · Zoubin Ghahramani · Katherine Heller · Tim Salimans · Max Welling · Matthew D. Hoffman -
2017 Workshop: Advances in Approximate Bayesian Inference »
Francisco Ruiz · Stephan Mandt · Cheng Zhang · James McInerney · James McInerney · Dustin Tran · Dustin Tran · David Blei · Max Welling · Tamara Broderick · Michalis Titsias -
2017 Poster: Hierarchical Implicit Models and Likelihood-Free Variational Inference »
Dustin Tran · Rajesh Ranganath · David Blei -
2017 Poster: Structured Embedding Models for Grouped Data »
Maja Rudolph · Francisco Ruiz · Susan Athey · David Blei -
2017 Poster: Variational Inference via $\chi$ Upper Bound Minimization »
Adji Bousso Dieng · Dustin Tran · Rajesh Ranganath · John Paisley · David Blei -
2017 Poster: Context Selection for Embedding Models »
Liping Liu · Francisco Ruiz · Susan Athey · David Blei -
2016 : Causal Inference for Recommendation Systems »
David Blei -
2016 : Panel Discussion »
Shakir Mohamed · David Blei · Ryan Adams · José Miguel Hernández-Lobato · Ian Goodfellow · Yarin Gal -
2016 : Deep exponential families »
David Blei -
2016 Workshop: Advances in Approximate Bayesian Inference »
Tamara Broderick · Stephan Mandt · James McInerney · Dustin Tran · David Blei · Kevin Murphy · Andrew Gelman · Michael I Jordan -
2016 Poster: Operator Variational Inference »
Rajesh Ranganath · Dustin Tran · Jaan Altosaar · David Blei -
2016 Poster: The Generalized Reparameterization Gradient »
Francisco Ruiz · Michalis Titsias · David Blei -
2016 Poster: Exponential Family Embeddings »
Maja Rudolph · Francisco Ruiz · Stephan Mandt · David Blei -
2016 Tutorial: Variational Inference: Foundations and Modern Methods »
David Blei · Shakir Mohamed · Rajesh Ranganath -
2015 Workshop: Advances in Approximate Bayesian Inference »
Dustin Tran · Tamara Broderick · Stephan Mandt · James McInerney · Shakir Mohamed · Alp Kucukelbir · Matthew D. Hoffman · Neil Lawrence · David Blei -
2015 Poster: The Population Posterior and Bayesian Modeling on Streams »
James McInerney · Rajesh Ranganath · David Blei -
2015 Poster: Automatic Variational Inference in Stan »
Alp Kucukelbir · Rajesh Ranganath · Andrew Gelman · David Blei -
2015 Spotlight: Automatic Variational Inference in Stan »
Alp Kucukelbir · Rajesh Ranganath · Andrew Gelman · David Blei -
2015 Poster: Copula variational inference »
Dustin Tran · David Blei · Edo M Airoldi