Timezone: »

 
Poster
Stochastic Stein Discrepancies
Jackson Gorham · Anant Raj · Lester Mackey

Thu Dec 10 09:00 PM -- 11:00 PM (PST) @ Poster Session 6 #1793
in Poster Session 7 »

Stein discrepancies (SDs) monitor convergence and non-convergence in approximate inference when exact integration and sampling are intractable. However, the computation of a Stein discrepancy can be prohibitive if the Stein operator -- often a sum over likelihood terms or potentials -- is expensive to evaluate. To address this deficiency, we show that stochastic Stein discrepancies (SSDs) based on subsampled approximations of the Stein operator inherit the convergence control properties of standard SDs with probability 1. Along the way, we establish the convergence of Stein variational gradient descent (SVGD) on unbounded domains, resolving an open question of Liu (2017). In our experiments with biased Markov chain Monte Carlo (MCMC) hyperparameter tuning, approximate MCMC sampler selection, and stochastic SVGD, SSDs deliver comparable inferences to standard SDs with orders of magnitude fewer likelihood evaluations.

Keywords: Supervised Deep Networks · Computer Vision · Applications · Algorithms -> Classification; Deep Learning; Deep Learning

Author Information

Jackson Gorham (Whisper.ai)
Anant Raj (Max Planck Institute for Intelligent Systems)
Lester Mackey (Microsoft Research)

More from the Same Authors