New arXiv paper proposes reduced cost influence functions for predict-then-optimize under noisy data
A new arXiv paper (2611.02086) introduces reduced cost influence functions for predict-then-optimize settings, aiming to keep ML models reliable in decision-making pipelines such as vehicle routing, dynamic pricing, and clinical decision support when data is noisy.
Read the full story at arXiv stat.ML ↗
Timeline · 1 report
- 2026-11-03 05:00 · arXiv stat.ML
Reduced Cost Influence Functions for Predict-then-Optimize under Noisy Data