Research map / Learning theory and algorithms / Prediction and decision losses
Obtaining Calibrated Probabilities with Personalized Ranking Models
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 Gaussian or Gamma calibration with UERM rather than naive log-loss fitting (ablation) | Parametric calibration of personalized ranking models · Yahoo!R3 and Coat | ECE: ECE decreases by 7.40%$\sim$76.52% | Helped |
| Use Gaussian or Gamma calibration with UERM rather than the best UERM Platt/Beta competitor | Personalized ranking models · Yahoo!R3 and Coat | ECE: 5.21%$\sim$25.85% | Helped |
| Allow class-conditional Gaussian variances to differ, rather than using Platt's equal-variance form | Gaussian calibration · Yahoo!R3 and Coat | ECE | Helped |
| Use a Gamma score-distribution calibration function that can capture skew | Gamma calibration · Yahoo!R3 and Coat | ECE | Helped |
| Use parametric calibration rather than non-parametric binning methods | Personalized ranking models · Yahoo!R3 and Coat | ECE | Helped |
| Replace the ranking score with a calibrated probability as a post-processing step | Pre-trained personalized ranking models · Ranking performance | ranking performance | No clear effect |
Closest papers by meaning
- Calibrated Preference Learning: The Case of Label Ranking (ICML 2026 · similarity 0.89)
- Calibration Matters: Tackling Maximization Bias in Large-scale Advertising Recommendation Systems (ICLR 2023 · similarity 0.85)
- Preference Diffusion for Recommendation (ICLR 2025 · similarity 0.84)
- From Pairwise to Ranking: Climbing the Ladder to Ideal Collaborative Filtering with Pseudo-Ranking (AAAI 2025 · similarity 0.83)
- Improving Model Probability Calibration by Integration of Large Data Sources with Biased Labels (AAAI 2025 · similarity 0.83)
- Adaptively Learning to Select-Rank in Online Platforms (ICML 2024 · similarity 0.83)
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.