Decision-Aware Training for Sample-Based Generative Models
Abstract
Sample-based generative models are increasingly used for probabilistic forecasting in high-stakes decision settings, yet their training objectives are blind to the decision maker's cost structure. These models are commonly trained with strictly proper scoring rules, such as the energy score, which allocate their training signal in proportion to data density, with no awareness of where forecast errors are most costly for downstream decisions. We therefore propose decision-aware training for sample-based generative models, augmenting the energy score objective with a differentiable decision loss that directly penalises the cost incurred by acting on the model's forecast. This combined loss is theoretically grounded, as the decision loss is itself a proper scoring rule. We compute the decision loss via a differentiable optimisation layer. Its gradient concentrates in cost-sensitive regions of the output space, making the method's effects interpretable and predictable from the cost structure. We validate the method on one synthetic and two real-world tasks. In the synthetic task, the method corrects the mode weights of a learned bimodal distribution; in a wind power dispatch task, it concentrates improvements in the rare but costly tail region, and in a frost protection task, it improves how well the decision costs are anticipated. Our method yields generative models that retain full probabilistic forecasts while being better aligned with the decision maker's specific cost structure.
Related
- What to Reuse in Generative Robot Policies? A Solver-Aware Mathematical Analysis of Training-Free Output Caching for Action Generation
- Near-Optimal Sample Complexity for Recursive Entropic Risk Reinforcement Learning with a Generative Model
- Sample-Efficient Generative Conformal Prediction
- PTED: A multi-dimensional two-sample test for scientific inference and generative machine learning
- Two-Sample Testing via Generative Processes
BibTeX
@misc{raeth2026decisionaware,
title = {{Decision-Aware Training for Sample-Based Generative Models}},
author = {Kornelius Raeth and Nicole Ludwig},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2607.01171},
eprint = {2607.01171},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2607.01171},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Decision-Aware Training for Sample-Based Generative Models
AU - Kornelius Raeth
AU - Nicole Ludwig
PY - 2026
DA - 2026/10/02
PB - arXiv
DO - 10.48550/arxiv.2607.01171
UR - https://arxiv.org/abs/2607.01171
AB - Sample-based generative models are increasingly used for probabilistic forecasting in high-stakes decision settings, yet their training objectives are blind to the decision maker's cost structure. These models are commonly trained with strictly proper scoring rules, such as the energy score, which allocate their training signal in proportion to data density, with no awareness of where forecast errors are most costly for downstream decisions. We therefore propose decision-aware training for sample-based generative models, augmenting the energy score objective with a differentiable decision loss that directly penalises the cost incurred by acting on the model's forecast. This combined loss is theoretically grounded, as the decision loss is itself a proper scoring rule. We compute the decision loss via a differentiable optimisation layer. Its gradient concentrates in cost-sensitive regions of the output space, making the method's effects interpretable and predictable from the cost structure. We validate the method on one synthetic and two real-world tasks. In the synthetic task, the method corrects the mode weights of a learned bimodal distribution; in a wind power dispatch task, it concentrates improvements in the rare but costly tail region, and in a frost protection task, it improves how well the decision costs are anticipated. Our method yields generative models that retain full probabilistic forecasts while being better aligned with the decision maker's specific cost structure.
ER -