Learning Best-in-Class Policies for the Predict-then-Optimize Framework

M Huang, V Gupta - arxiv preprint arxiv:2402.03256, 2024 - arxiv.org
We propose a novel family of decision-aware surrogate losses, called Perturbation Gradient
(PG) losses, for the predict-then-optimize framework. These losses directly approximate the …

[PDF][PDF] Decision-Focused Learning with Directional Gradients

V Gupta, M Huang - Training, 2024 - faculty.marshall.usc.edu
We propose a novel family of decision-aware surrogate losses, called Perturbation Gradient
(PG) losses, for the predict-then-optimize framework. The key idea is to connect the …

Decision-Focused Learning with Directional Gradients

M Huang, V Gupta - The Thirty-eighth Annual Conference on Neural …, 2024 - openreview.net
We propose a novel family of decision-aware surrogate losses, called Perturbation Gradient
(PG) losses, for the predict-then-optimize framework. These losses directly approximate the …

More Data or Better Data? Impact of Costly Data Collection on the Newsvendor Problem

Z Zhang, HS Ahn, L Baardman - Impact of Costly Data Collection on …, 2024 - papers.ssrn.com
Data-driven algorithms and policies have flourished in many areas of operations
management. While many applications can utilize preexisting data, there are many cases …