no code implementations • 11 May 2023 • Mo Liu, Paul Grigas, Heyuan Liu, Zuo-Jun Max Shen
We develop the first active learning method in the predict-then-optimize framework.
no code implementations • 15 Jun 2022 • Heyuan Liu, Paul Grigas
We propose an algorithm that mixes a prediction step based on the "Smart Predict-then-Optimize (SPO)" method with a dual update step based on mirror descent.
no code implementations • NeurIPS 2021 • Heyuan Liu, Paul Grigas
We develop risk bounds and uniform calibration results for the SPO+ loss relative to the SPO loss, which provide a quantitative way to transfer the excess surrogate risk to excess true risk.