Research map / Learning theory and algorithms / Prediction and decision losses
Decision-Theoretic Approaches for Improved Learning-Augmented Algorithms
Spyros Angelopoulos, Christoph Dürr, Georgii Melidi
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use distance-based and CVaR thresholds instead of the BP_rho family | learning-augmented ski rental · Ski rental (b=10, r=5, delta=0.9) | average performance ratio: 1.677 -> 1.337 | Improved |
| increase CVaR risk-aversion alpha from 0.1 to 0.9 (ablation) | CVaR algorithm, ski rental · Ski rental (b=10, r=5, delta=0.9) | average performance ratio: 1.340 -> 1.367 | Hurt |
| increase CVaR risk-aversion alpha from 0.1 to 0.9 (ablation) | CVaR algorithm, one-max search · One-max search (M=1000, r=100, delta=0.9) | average performance ratio: 9.771 -> 6.022 | Improved |
| increase CVaR risk-aversion alpha from 0.1 to 0.9 (ablation) | CVaR algorithm, one-max search · One-max search (M=1000, r=100, delta=0.9) | expected profit: 35.795 -> 27.500 | Hurt |
| use Max or Avg distance thresholds instead of Pareto-optimal PO1 and delta-Tol | one-max search · One-max search (M=1000, r=100, delta=0.9) | average performance ratio: 4.630 -> 4.394 | Improved |
| use CVaR thresholds instead of delta-Tol | one-max search · One-max search (M=1000, r=100, delta=0.9) | expected profit: 5.475 -> 27.500 | Improved |
| use CVaR thresholds instead of PO1 | one-max search · One-max search (M=1000, r=100, delta=0.9) | average performance ratio: 4.630 -> 6.022 | Hurt |
| shrink the prediction range or concentrate the weight function around y (ablation) | Max, Avg, CVaR algorithms · appendix parameter study | performance | Improved |
Closest papers by meaning
- Online Algorithms with Uncertainty-Quantified Predictions (ICML 2024 · similarity 0.90)
- Learning-Augmented Algorithms with Explicit Predictors (NeurIPS 2024 · similarity 0.87)
- Learning-Augmented Algorithms for MTS with Bandit Access to Multiple Predictors (ICML 2025 · similarity 0.87)
- Learning Augmented Exact Exponential Algorithms (NeurIPS 2026 · similarity 0.87)
- Learning-Augmented Online Bidding in Stochastic Settings (NeurIPS 2025 · similarity 0.87)
- Pareto-Optimality, Smoothness, and Stochasticity in Learning-Augmented One-Max-Search (ICML 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
