DFF: Decision-Focused Fine-tuning for Smarter Predict-then-Optimize with Limited Data

2025-01-03

AI Summary

To address decision performance degradation in prediction-and-optimization (PO) under data scarcity, this paper proposes a Decision-Driven Fine-tuning (DFF) framework. DFF introduces a trust-region-constrained bias correction module that minimizes end-to-end decision loss without compromising the physical interpretability of the original predictive model. It is the first work to embed decision-focused learning into the fine-tuning paradigm; we theoretically prove that the corrected bias is strictly bounded. The framework supports non-differentiable and black-box predictors while balancing decision improvement and prediction reliability. Evaluated on real-world tasks—including network flow scheduling, portfolio optimization, and resource allocation—DFF reduces average decision loss by 18.7% and prediction shift by 63%, demonstrating strong generalizability and strict adherence to fine-tuning constraints.

Technology Category

• Search and Optimization: Learning to Search
• Reasoning under Uncertainty: Stochastic Optimization
• Constraint Satisfaction and Optimization: Constraint Learning and Acquisition

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Decision-focused learning (DFL) offers an end-to-end approach to the predict-then-optimize (PO) framework by training predictive models directly on decision loss (DL), enhancing decision-making performance within PO contexts. However, the implementation of DFL poses distinct challenges. Primarily, DL can result in deviation from the physical significance of the predictions under limited data. Additionally, some predictive models are non-differentiable or black-box, which cannot be adjusted using gradient-based methods. To tackle the above challenges, we propose a novel framework, Decision-Focused Fine-tuning (DFF), which embeds the DFL module into the PO pipeline via a novel bias correction module. DFF is formulated as a constrained optimization problem that maintains the proximity of the DL-enhanced model to the original predictive model within a defined trust region. We theoretically prove that DFF strictly confines prediction bias within a predetermined upper bound, even with limited datasets, thereby substantially reducing prediction shifts caused by DL under limited data. Furthermore, the bias correction module can be integrated into diverse predictive models, enhancing adaptability to a broad range of PO tasks. Extensive evaluations on synthetic and real-world datasets, including network flow, portfolio optimization, and resource allocation problems with different predictive models, demonstrate that DFF not only improves decision performance but also adheres to fine-tuning constraints, showcasing robust adaptability across various scenarios.

Problem

• Limited Data
• Prediction Accuracy
• Decision Quality

Innovation

• Decision-Focused Fine-tuning
• Predictive Accuracy
• Limited Data