Research map / Learning theory and algorithms / Prediction and decision losses
Generalized Results for the Existence and Consistency of the MLE in the Bradley-Terry-Luce Model
Heejong Bong, Alessandro Rinaldo
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 the smallest nonzero eigenvalue of Fisher information to state the MLE existence condition | Bradley-Terry-Luce MLE · Arbitrary connected comparison graph | MLE existence failure probability: 2/√d | Improved |
| Use proxy function h_d and Fisher information instead of requiring bounded dynamic range | Bradley-Terry-Luce MLE · Arbitrary comparison graph | squared proxy ℓ₂ error bound: c₁ (κ/λ₂(ℐ*)) td/n | Improved |
| Jointly account for graph topology and score gaps through Fisher information | Bradley-Terry-Luce MLE · Banded comparison graph | squared error bound dependence on B: 2c'' td³/(TW³) e^(4B) -> c₁ td³/(TW³) e^(4BW/d) | Improved |
| Use the general proxy-function bound on Erdős–Rényi graphs | Bradley-Terry-Luce MLE · Erdős–Rényi graph | squared error rate | No clear effect |
| Use a proxy-function MLE bound without known dynamic-range bound B | Bradley-Terry-Luce MLE · Arbitrary graph with bounded dynamic range | squared L-norm error bound: c₁ (e^(−B)+e^B)^4 td/n; c'' (e^(−B)+e^B)^4 td/n | No clear effect |
| Use Fisher-information bound rather than the previous bound | Bradley-Terry-Luce MLE · Simulated complete, star, and barbell graphs | error-rate growth as dynamic range B varies | Improved |
| Use Fisher-information bound rather than the previous bound | Bradley-Terry-Luce MLE · Simulated banded comparison graphs | error-rate growth as comparison width W varies | Improved |
Closest papers by meaning
- Estimation from Pairwise Comparisons: Sharp Minimax Bounds with Topology Dependence (JMLR 2016 · similarity 0.89)
- Dynamic Ranking with the BTL Model: A Nearest Neighbor based Rank Centrality Method (JMLR 2023 · similarity 0.89)
- Efficient Computation of Rankings from Pairwise Comparisons (JMLR 2023 · similarity 0.89)
- Generalized Bradley-Terry Models for Score Estimation from Paired Comparisons (AAAI 2024 · similarity 0.89)
- An Analysis of Elo Rating Systems via Markov Chains (NeurIPS 2024 · similarity 0.86)
- Active Ranking and Matchmaking, with Perfect Matchings (ICML 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.
