Decision-Focused Learning in Network Interdiction Games

2026-08-09

AI Summary

This work addresses the performance degradation of conventional decision-focused learning (DFL) in shortest-path network interdiction games, where cost estimators suffer from detrimental decision equivalence classes, sometimes rendering DFL inferior to prediction-focused approaches. To overcome this limitation, the paper proposes adversarial decision-focused learning (A-DFL), which integrates an adversarial training mechanism that replaces original samples with perturbed interdiction scenarios to effectively eliminate harmful decision equivalence classes. By combining Stackelberg game modeling, machine learning predictors, and adversarial strategies, A-DFL enables end-to-end robust decision optimization. Experimental results on both synthetic and real-world networks demonstrate that A-DFL significantly outperforms standard DFL and prediction-focused baselines, successfully restoring and enhancing the optimization efficacy of the decision-focused paradigm.

Abstract

We study decision-focused learning (DFL) in shortest-path network interdiction (SPNI) games, a Stackelberg game where an interdictor (leader) strengthens the networks' arcs against attacks, while an evader (follower) who is uncertain about costs of attacking network arcs relies on a machine-learned predictor to identify the shortest path. While DFL is highly effective as an end-to-end optimization framework, we show that it faces a fundamental structural failure when employed in this game setting: its training objective admits a broad decision-equivalence class of cost estimators that achieve zero nominal loss yet fail under interdiction, reversing DFL's usual advantage over a naive prediction-focused learning (PFL) approach. To address this, we propose Adversarial DFL (A-DFL), which replaces nominal training samples with interdicted scenarios to collapse the harmful equivalence class. Experiments on synthetic and real-world networks confirm that A-DFL restores DFL's advantage in this game setting, enabling effective end-to-end optimization.

Problem

• Decision-Focused Learning
• Network Interdiction
• Stackelberg Game
• Shortest-Path
• Structural Failure

Innovation

• Decision-Focused Learning
• Network Interdiction
• Adversarial Training
• Stackelberg Game
• End-to-End Optimization