Timezone: »

 
Oral
REBAR: Low-variance, unbiased gradient estimates for discrete latent variable models
George Tucker · Andriy Mnih · Chris J Maddison · John Lawson · Jascha Sohl-Dickstein

Wed Dec 06 10:35 AM -- 10:50 AM (PST) @ Hall A

Learning in models with discrete latent variables is challenging due to high variance gradient estimators. Generally, approaches have relied on control variates to reduce the variance of the REINFORCE estimator. Recent work \citep{jang2016categorical, maddison2016concrete} has taken a different approach, introducing a continuous relaxation of discrete variables to produce low-variance, but biased, gradient estimates. In this work, we combine the two approaches through a novel control variate that produces low-variance, \emph{unbiased} gradient estimates. Then, we introduce a novel continuous relaxation and show that the tightness of the relaxation can be adapted online, removing it as a hyperparameter. We show state-of-the-art variance reduction on several benchmark generative modeling tasks, generally leading to faster convergence to a better final log likelihood.

Author Information

George Tucker (Google Brain)
Andriy Mnih (DeepMind)
Chris J Maddison (University of Oxford / DeepMind)
Dieterich Lawson (Google Brain)
Jascha Sohl-Dickstein (Google Brain)

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors