Online learning with optimism and delay
Files
Published version
Date
2021
DOI
Authors
Flaspohler, Genevieve
Orabona, Francesco
Cohen, Judah
Mouatadid, Soukayna
Oprescu, Miruna
Orenstein, Paulo
Mackey, Lester
Version
Published version
OA Version
Citation
G. Flaspohler, F. Orabona, J. Cohen, S. Mouatadid, M. Oprescu, P. Orenstein, L. Mackey. 2021. "Online Learning with Optimism and Delay." INTERNATIONAL CONFERENCE ON MACHINE LEARNING, VOL 139.
Abstract
Inspired by the demands of real-time climate and
weather forecasting, we develop optimistic online
learning algorithms that require no parameter
tuning and have optimal regret guarantees under
delayed feedback. Our algorithms—DORM,
DORM+, and AdaHedgeD—arise from a novel
reduction of delayed online learning to optimistic
online learning that reveals how optimistic hints
can mitigate the regret penalty caused by delay.
We pair this delay-as-optimism perspective with
a new analysis of optimistic learning that exposes
its robustness to hinting errors and a new metaalgorithm
for learning effective hinting strategies
in the presence of delay. We conclude by benchmarking
our algorithms on four subseasonal climate
forecasting tasks, demonstrating low regret
relative to state-of-the-art forecasting models.