Research map / Learning theory and algorithms / Prediction and decision losses
Decision-Focused Learning: Through the Lens of Learning to Rank
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use listwise ranking loss instead of Pointwise on higher-degree shortest-path tasks | Decision-focused learning · Shortest Path Deg-4, Deg-6, and Deg-8 | percentage regret: 19.27% -> 7.71%; 72.18% -> 8.29%; 178.16% -> 12.38% | Helped |
| Use pairwise-difference loss instead of Pointwise | Decision-focused learning · Matching-1 and Matching-2 | percentage regret: 90.17% -> 88.23%; 89.86% -> 85.87% | Helped |
| Use pairwise-difference loss instead of Pointwise | Decision-focused learning · Energy-1 and Energy-2 | percentage regret: 2.51% -> 1.63%; 2.83% -> 2.04% | Helped |
| Use pointwise loss rather than two-stage MSE training | Decision-focused learning · All reported problem instances | percentage regret | Hurt |
| Reduce solution-oracle solve probability from 100% to 10% (ablation) | Ranking losses · Energy-1 | per-epoch training time: 110 -> 30; 110 -> 30; 110 -> 30; 100 -> 25 | Helped |
| Reduce solution-oracle solve probability from 100% to 10% (ablation) | Ranking losses · Energy-2 | per-epoch training time: 130 -> 40; 140 -> 45; 140 -> 45; 120 -> 40 | Helped |
| Reduce solution-oracle solve probability from 100% to 10% (ablation) | Ranking losses except Pointwise · Energy-1 and Energy-2 | percentage regret: Energy-1 Pairwise 1.64% -> 1.65%; Pairwise-diff 1.63% -> 1.61%; Listwise 1.72% -> 1.67%; Energy-2 Pairwise 2.05% -> 2.08 | No clear effect |
| Combine listwise loss with MSE using a positive listwise-loss weight (ablation) | Listwise ranking training · Energy-1 and Energy-2 | percentage regret | Mixed |
| Replace cross-entropy listwise loss with KL-divergence loss (ablation) | Listwise ranking · Experiments reported in text | regret | No clear effect |
Closest papers by meaning
- Solver-Free Decision-Focused Learning for Linear Optimization Problems (NeurIPS 2025 · similarity 0.90)
- Leaving the Nest: Going beyond Local Loss Functions for Predict-Then-Optimize (AAAI 2024 · similarity 0.87)
- Feasibility-Aware Decision-Focused Learning for Predicting Parameters in the Constraints (NeurIPS 2025 · similarity 0.86)
- A Generalization Result for Convergence in Learning-to-Optimize (ICML 2025 · similarity 0.85)
- What Data Enables Optimal Decisions? An Exact Characterization for Linear Optimization (NeurIPS 2025 · similarity 0.85)
- Locally Convex Global Loss Network for Decision-Focused Learning (AAAI 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.