Learning Scenario Reduction for Two-Stage Robust Optimization with Discrete Uncertainty

PRISE uses sequential lookahead to select representative scenarios for two-stage robust optimization, while NeurPRISE learns a GNN-Transformer surrogate via imitation learning that achieves 7-200x speedups and strong zero-shot generalization.

TIANJUE LIN, Jianan Zhou, Jieyi Bi, Yaoxin Wu and 3 more

Sydney Poster Session 4, Wed, Dec 9, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026