Research map / Learning theory and algorithms / Prediction and decision losses
A New and Flexible Approach to the Analysis of Paired Comparison Data
Ivo F. D. Oliveira, Nir Ailon, Ori Davidov
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 |
|---|---|---|---|
| add PolyRank refinement to the unrefined estimator | pairwise probability estimate · Simulation, I=20, 50 comparisons | L2 distance to P* | Improved |
| use degree D=7 instead of D=3 or 5 (ablation) | PolyRank · Round-robin, increasing items | ||P*-P||2/I2 | Hurt |
| use under-fitted D=3 instead of true D=5 (ablation) | PolyRank · Round-robin, increasing items | ||P*-P||2/I2 | Improved |
| increase matches per pair | PolyRank · Round-robin, I=10 | estimator variance | Improved |
| increase mij from 1 to 3 (I=6) (ablation) | AIC degree selection · Degree selection I=6 | frequency selecting D=3: 0% -> 69.8% | Improved |
| use AIC instead of cross-validation | degree selection · Degree selection | selection accuracy | No clear effect |
| use BIC instead of AIC (I=12, mij=3) | degree selection · Degree selection I=12 | frequency selecting D=5: 12.2% -> 0.7% | Hurt |
| use polynomial comparison function instead of BTL | comparison function · Computer chess, p_ij=p_jk=0.9 | predicted p_ik: 0.99 vs 0.92 | Mixed |
| increase polynomial degree D beyond 7 | estimated F · Computer chess, first 100 engines | shape of F | No clear effect |
| increase number of engines I to 60 or more | estimated F · Computer chess, D=7 | shape of F | No clear effect |
Closest papers by meaning
- Estimation from Pairwise Comparisons: Sharp Minimax Bounds with Topology Dependence (JMLR 2016 · similarity 0.89)
- Generalized Results for the Existence and Consistency of the MLE in the Bradley-Terry-Luce Model (ICML 2022 · similarity 0.89)
- Pairwise Comparisons without Stochastic Transitivity: Model, Theory and Applications (JMLR 2026 · similarity 0.88)
- Two-Sample Testing on Ranked Preference Data and the Role of Modeling Assumptions (JMLR 2022 · similarity 0.86)
- Active Ranking and Matchmaking, with Perfect Matchings (ICML 2024 · similarity 0.86)
- Aggregating Quantitative Relative Judgments: From Social Choice to Ranking Prediction (NeurIPS 2024 · 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.
