Research map / Learning theory and algorithms / Prediction and decision losses
Least Square Calibration for Peer Reviews
Sijun Tan, Jibang Wu, Xiaohui Bei, Haifeng Xu
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 linear-constrained LSC instead of averaging scores (ablation) | reviewer calibration · Synthetic noiseless | precision (%): 40.0 ± 4.0 -> 100 ± 0 | Improved |
| use LSC (linear) instead of QP (ablation) | reviewer calibration · Synthetic noisy | precision (%): 69.2 ± 4.6 -> 80.1 ± 2.9 | Improved |
| use LSC (linear) instead of Bayesian (ablation) | reviewer calibration · Synthetic noisy | precision (%): 71.5 ± 3.1 -> 80.1 ± 2.9 | Improved |
| use monotone instead of linear constraint (ablation) | LSC · Synthetic linear noiseless | precision (%): 100 ± 0 -> 91.7 ± 1.8 | Hurt |
| use monotone instead of linear constraint (ablation) | LSC · Synthetic linear noisy | precision (%): 80.1 ± 2.9 -> 75.9 ± 2.6 | Hurt |
| QP baseline under noise (ablation) | QP · Synthetic linear | precision (%): 97.1 ± 1.7 -> 69.2 ± 4.6 | Hurt |
| give LSC exact per-reviewer function type (mixed) (ablation) | LSC · Mixed scoring noiseless | precision (%): 76.4 ± 3.8 -> 93.2 ± 2.5 | Improved |
| give LSC exact per-reviewer function type (mixed) (ablation) | LSC · Mixed scoring noisy | precision (%): 67.7 ± 3.1 -> 71.1 ± 3.4 | Improved |
| use LSC (linear) when all reviewers are random monotone (ablation) | LSC · Synthetic noiseless | precision: 4% behind LSC (mono) | Hurt |
| use LSC (linear) on real peer-grading data (ablation) | LSC · Peer-Grading Dataset | precision (%): 80.9 -> 82.2 | Improved |
| use LSC (mono) on real peer-grading data (ablation) | LSC · Peer-Grading Dataset | average gap: 0.48 -> 0.55 | Hurt |
| increase papers per reviewer k from 3 to 7 (ablation) | LSC · Synthetic | precision | Improved |
Closest papers by meaning
- From Replication to Redesign: Exploring Pairwise Comparisons for LLM-Based Peer Review (NeurIPS 2025 · similarity 0.82)
- Eliciting Honest Information from Authors Using Sequential Review (AAAI 2024 · similarity 0.82)
- Position: The AI Conference Peer Review Crisis Demands Author Feedback and Reviewer Rewards (ICML 2025 · similarity 0.81)
- PRISMM-Bench: A Benchmark of Peer-Review Grounded Multimodal Inconsistencies (ICLR 2026 · similarity 0.81)
- Allocate Marginal Reviews to Borderline Papers Using LLM Comparative Ranking (NeurIPS 2026 · similarity 0.81)
- PiCO: Peer Review in LLMs based on Consistency Optimization (ICLR 2025 · similarity 0.80)
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.
