Low-variance Black-box Gradient Estimates for the Plackett-Luce Distribution

  1. Photo Artyom Gadetsky Artyom Gadetsky
  2. Photo Kirill Struminsky Kirill Struminsky
  3. Photo Christopher Robinson Christopher Robinson
  4. Photo Novi Quadrianto Novi Quadrianto
  5. Photo Dmitry Vetrov Dmitry Vetrov

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.