Research map / Learning theory and algorithms / Submodular and game algorithms
From Social to Individuals: A Parsimonious Path of Multi-Level Models for Crowdsourced Preference Aggregation
Qianqian Xu, Jiechao Xiong, Xiaochun Cao, Qingming Huang, Yuan Yao
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 |
|---|---|---|---|
| replace HodgeRank with the multi-level mixed-effects model (linear loss) | HodgeRank · Simulated data | mean test error (mismatch ratio): 0.3097 -> 0.1892 | Improved |
| use Bradley-Terry loss instead of linear loss | Mixed-effects LBI · Simulated data | mean test error (mismatch ratio): 0.1892 -> 0.1799 | Improved |
| use Thurstone-Mosteller loss instead of linear loss | Mixed-effects LBI · Simulated data | mean test error (mismatch ratio): 0.1892 -> 0.1822 | Improved |
| replace HodgeRank with the multi-level model (Bradley-Terry loss) | HodgeRank · MovieLens 1M subset | mean test error (mismatch ratio): 0.4361 -> 0.3278 | Improved |
| use Bradley-Terry loss instead of linear loss | Mixed-effects LBI · MovieLens 1M subset | mean test error (mismatch ratio): 0.3316 -> 0.3278 | Improved |
| replace HodgeRank with the multi-level model (Bradley-Terry loss) | HodgeRank · IQA (reference image 1) | mean test error (mismatch ratio): 0.1874 -> 0.0708 | Improved |
| use Bradley-Terry loss instead of linear loss | Mixed-effects LBI · IQA (reference image 1) | mean test error (mismatch ratio): 0.1094 -> 0.0708 | Improved |
| use Thurstone-Mosteller loss instead of linear loss | Mixed-effects LBI · IQA (reference image 1) | mean test error (mismatch ratio): 0.1094 -> 0.0731 | Improved |
| replace HodgeRank with the multi-level model (Bradley-Terry loss) | HodgeRank · WorldCollege ranking | mean test error (mismatch ratio): 0.3108 -> 0.2555 | Improved |
| use Bradley-Terry loss instead of linear loss | Mixed-effects LBI · WorldCollege ranking | mean test error (mismatch ratio): 0.2670 -> 0.2555 | Improved |
| replace the best ranking baseline (URLR) with the multi-level model (linear loss) | URLR · WorldCollege ranking | mean test error (mismatch ratio): 0.3136 -> 0.2670 | Improved |
| increase threads from 1 to 16 in synchronized parallel LBI | SynPar-LBI · Simulated data | speedup: nearly linear speedup | Improved |
Closest papers by meaning
- Selective Preference Aggregation (ICML 2025 · similarity 0.84)
- Leveraging Sparsity for Sample-Efficient Preference Learning: A Theoretical Perspective (ICML 2025 · similarity 0.83)
- Impacts of Aggregation on Model Diversity and Consumer Utility (NeurIPS 2026 · similarity 0.83)
- Diverging Preferences: When do Annotators Disagree and do Models Know? (ICML 2025 · similarity 0.83)
- Preference Diffusion for Recommendation (ICLR 2025 · similarity 0.83)
- Enhancing Preference-based Linear Bandits via Human Response Time (NeurIPS 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.
