1 paper
Javier Burroni, Kenta Takatsu, Justin Domke +1
We propose the use of U-statistics to reduce variance for gradient estimation in importance-weighted variational inference. The key observation is that, given a base gradient estim…