Timezone: »

Streaming Variational Bayes
Tamara Broderick · Nicholas Boyd · Andre Wibisono · Ashia C Wilson · Michael Jordan

Thu Dec 05 07:00 PM -- 11:59 PM (PST) @ Harrah's Special Events Center, 2nd Floor

We present SDA-Bayes, a framework for (S)treaming, (D)istributed, (A)synchronous computation of a Bayesian posterior. The framework makes streaming updates to the estimated posterior according to a user-specified approximation primitive function. We demonstrate the usefulness of our framework, with variational Bayes (VB) as the primitive, by fitting the latent Dirichlet allocation model to two large-scale document collections. We demonstrate the advantages of our algorithm over stochastic variational inference (SVI), both in the single-pass setting SVI was designed for and in the streaming setting, to which SVI does not apply.

Author Information

Tamara Broderick (MIT)
Nicholas Boyd (UC Berkeley)
Andre Wibisono (Georgia Tech)
Ashia C Wilson (UC Berkeley)
Michael Jordan (UC Berkeley)

More from the Same Authors