Online Learning with Optimism and Delay
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 meta-algorithm 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.
Cite
@article{arxiv.2106.06885,
title = {Online Learning with Optimism and Delay},
author = {Genevieve Flaspohler and Francesco Orabona and Judah Cohen and Soukayna Mouatadid and Miruna Oprescu and Paulo Orenstein and Lester Mackey},
journal= {arXiv preprint arXiv:2106.06885},
year = {2021}
}
Comments
ICML 2021. 9 pages of main paper and 26 pages of appendix text