Low-Variance Black-Box Gradient Estimates for the Plackett-Luce Distribution

  • Artyom Gadetsky National Research University Higher School of Economics
  • Kirill Struminsky National Research University Higher School of Economics
  • Christopher Robinson University of Sussex
  • Novi Quadrianto University of Sussex
  • Dmitry Vetrov National Research University Higher School of Economics

Abstract

Learning models with discrete latent variables using stochastic gradient descent remains a challenge due to the high variance of gradient estimates. Modern variance reduction techniques mostly consider categorical distributions and have limited applicability when the number of possible outcomes becomes large. In this work, we consider models with latent permutations and propose control variates for the Plackett-Luce distribution. In particular, the control variates allow us to optimize black-box functions over permutations using stochastic gradient descent. To illustrate the approach, we consider a variety of causal structure learning tasks for continuous and discrete data. We show that our method outperforms competitive relaxation-based optimization methods and is also applicable to non-differentiable score functions.

Published
2020-04-03
Section
AAAI Technical Track: Reasoning under Uncertainty