# Reducing the variance in online optimization by transporting past gradients

NeurIPS 2019 (2019) (to appear)