Skip to content

Latest commit

 

History

History
24 lines (19 loc) · 923 Bytes

BleKucMca16.md

File metadata and controls

24 lines (19 loc) · 923 Bytes

Variational Inference: A Review for Statisticians

D. Blei, A. Kucukelbir, J. McAuliffe. Variational Inference: A Review for Statisticians. 2016.

tl;dr

  • Takeaway 1
  • Takeaway 2
  • Open question or critique

Thoughts

What is this paper showing? What are the contributions? What kind of paper is it: theoretical, applied, something in the middle? Where does it fit into the prior body of work?

Outline

  • Basics
    • Variational inference is an optimization problem (vs sampling in MCMC) using KL divergence
    • Pros: VI better on large datasets, faster
    • Cons: no guarantees on variance, not exact samples
  • Simplest example
    • "evidence" is demominator of p(z|x), aka p(x), exponential to calc
    • evidence lower bound (ELBO) allows us to get around difficulty calculating p(x)
    • mean-field assumes all latent variables are mutually independent

Q's for authors

TODO