“Edward”: A library for probabilistic modeling, inference, and criticism
Abstract https://www.meetup.com/NYC-Machine-Learning/events/236943279/
http://www.zentation.com/viewer2/webcast/NAPNgdDUBF/Dustin-Tran—Spotify-Talk-(2017-01-19)
Probabilistic modeling is a powerful approach for analyzing empirical information. In this talk, I will provide an overview of Edward, a software library for probabilistic modeling. Formally, Edward is a probabilistic programming system built on computational graphs,supporting compositions of both models and inference for flexible experimentation. For example, Edward makes it easy to fit the same model using a variety of composable inferences, ranging from point estimation, to variational inference, to MCMC. Edward is also integrated into TensorFlow, providing significant speedups over existing probabilistic systems. As examples, I will show how Edward can be leveraged for expanding the frontier of variational inference and deep generative models.
Bio
Dustin Tran is a Ph.D. student in Computer Science at Columbia, where he is advised by David Blei and Andrew Gelman. He works in the fields of Bayesian statistics, machine learning, and deep learning, and in particular has made recent advances in variational inference. He is a core developer for Edward, a library for probabilistic modeling,inference, and criticism. He is also a member of the Stan development team and collaborates with Google Brain. Previously, he was a Statistics Ph.D. student at Harvard before transferring to Columbia.