Research map / Learning theory and algorithms / Prediction and decision losses

A Little Robustness Is All You Need: Leveraging Predictions for Contextual Optimization

NeurIPS 2026 · Learning theory and algorithms › Prediction and decision losses · approach: Losses for decision-making

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Shows that follow-the-prediction policies for contextual optimization can be brittle, and proposes a simple robustified policy that wraps any black-box predictor.

Problem

Contextual optimization typically trains a predictor of the unknown state from context and then acts optimally for the predicted state; this follow-the-prediction policy can incur constant excess loss even when the predictor has arbitrarily small RMSE.

Approach

Robustify any black-box predictor's follow-the-prediction policy with a simple modification that hedges against prediction error, and show that this removes the constant excess-loss failure mode for several standard problems.

Key results

• Shows that even a minimum-variance predictor with arbitrarily small RMSE can incur constant excess loss under follow-the-prediction for several standard problems.

Surprising findings

• Follow-the-prediction can incur constant excess loss even when the predictor's RMSE is arbitrarily small.

Limitations

• Full text unavailable; results beyond the abstract could not be verified.

Experimental setup

Metrics: excess loss RMSE

Baselines: follow-the-prediction policy

Builds on: predict-then-optimize follow-the-prediction policy

Closest papers by meaning

• Contextual Online Pricing with (Biased) Offline Data (NeurIPS 2025 · similarity 0.88)
• Optimal Contextual Pricing under Agnostic Non-Lipschitz Demand (NeurIPS 2026 · similarity 0.88)
• Improved Algorithms for Contextual Dynamic Pricing (NeurIPS 2024 · similarity 0.88)
• Contextual Dynamic Pricing with Heterogeneous Buyers (NeurIPS 2025 · similarity 0.86)
• Learning to price with resource constraints: from full information to machine-learned prices (NeurIPS 2025 · similarity 0.86)
• Contextual Optimization Under Model Misspecification: A Tractable and Generalizable Approach (ICML 2025 · similarity 0.86)