REVIEW 5 cited by
Decision-Focused Learning: Foundations, State of the Art, Benchmark and Future Opportunities
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
Signed reviews
read the original abstract
Decision-focused learning (DFL) is an emerging paradigm that integrates machine learning (ML) and constrained optimization to enhance decision quality by training ML models in an end-to-end system. This approach shows significant potential to revolutionize combinatorial decision-making in real-world applications that operate under uncertainty, where estimating unknown parameters within decision models is a major challenge. This paper presents a comprehensive review of DFL, providing an in-depth analysis of both gradient-based and gradient-free techniques used to combine ML and constrained optimization. It evaluates the strengths and limitations of these techniques and includes an extensive empirical evaluation of eleven methods across seven problems. The survey also offers insights into recent advancements and future research directions in DFL. Code and benchmark: https://github.com/PredOpt/predopt-benchmarks
Forward citations
Cited by 5 Pith papers
-
DFF: Decision-Focused Fine-tuning for Smarter Predict-then-Optimize with Limited Data
Decision-Focused Fine-tuning corrects any backbone predictor's outputs within a bounded trust region using a residual scaling layer, improving decision regret on predict-then-optimize tasks.
-
Uplift modeling with continuous treatments: A predict-then-optimize approach
A predict-then-optimize framework that estimates dose-response curves and allocates continuous treatments via integer linear programming, with fairness and cost-sensitive objectives.
-
DistrictNet: Decision-aware learning for geographical districting
DistrictNet learns CMST edge weights with a GNN in a decision-aware way and produces districting solutions that are about 10% cheaper on average than existing methods on real cities.
-
Prediction Loss Guided Decision-Focused Learning
A gradient-perturbation method that combines prediction-loss and decision-loss gradients, with a decaying weight, to stabilize decision-focused learning.
-
Decision-Focused Learning for Complex System Identification: HVAC Management System Application
Decision-focused learning identifies RC model parameters inside a convex HVAC scheduling policy, reducing day-ahead cost prediction error from 389 to 16 euros in a 15-zone building simulation.
Discussion (0). Continue with ORCID to comment.