Research map / Large language models / Alignment and preferences
RankList – a Listwise Preference Learning Framework for Predicting Subjective Preferences
Abinay Reddy Naini, Fernando Diaz, Carlos Busso
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 log-sum-exp approximation instead of exact sum of pairwise log losses (ablation) | RankList · MSP-Podcast arousal | Kendall's Tau: 0.564 -> 0.591 | Improved |
| add log-sum-exp approximation instead of exact sum of pairwise log losses (ablation) | RankList · MSP-Podcast dominance | Kendall's Tau: 0.430 -> 0.461 | Improved |
| add skip-term comparisons to adjacent-only loss (ablation) | RankList · MSP-Podcast arousal | Kendall's Tau: 0.571 -> 0.591 | Improved |
| add skip-term comparisons to adjacent-only loss (ablation) | RankList · MSP-Podcast dominance | Kendall's Tau: 0.442 -> 0.461 | Improved |
| extend skip comparisons from skip-2 to skip-3 (ablation) | RankList · MSP-Podcast arousal | Kendall's Tau: 0.591 -> 0.580 | No clear effect |
| reduce skip comparisons from skip-2 to skip-1 (ablation) | RankList · MSP-Podcast arousal | Kendall's Tau: 0.591 -> 0.586 | No clear effect |
| add RankNet pairwise pre-training before listwise loss (ablation) | RankList · MSP-Podcast dominance | Kendall's Tau: 0.439 -> 0.461 | Improved |
| add RankNet pairwise pre-training before listwise loss (ablation) | RankList · MSP-Podcast arousal | Kendall's Tau: 0.582 -> 0.591 | Improved |
| use listwise RankList instead of pairwise RankNet | RankNet · MSP-Podcast arousal | Kendall's Tau: 0.526 -> 0.591 | Improved |
| use RankList instead of RnC on IEMOCAP dominance within-corpus | RnC · IEMOCAP dominance (within-corpus) | Kendall's Tau: 0.370 -> 0.372 | No clear effect |
| vary minimum score margin W for list construction (ablation) | RankList · MSP-Podcast | Kendall's Tau | Mixed |
| increase training size with matched visible comparisons, listwise versus pairwise | RankList · MSP-Podcast arousal | Kendall's Tau | Improved |
Closest papers by meaning
- PreferThinker: Reasoning-based Personalized Image Preference Assessment (ICLR 2026 · similarity 0.88)
- Learning Multi-Dimensional Human Preference for Text-to-Image Generation (CVPR 2024 · similarity 0.88)
- Diffusion Model as a Noise-Aware Latent Reward Model for Step-Level Preference Optimization (NeurIPS 2025 · similarity 0.87)
- Aligning Vision Models with Human Aesthetics in Retrieval: Benchmarks and Algorithms (NeurIPS 2024 · similarity 0.86)
- UniAR: A Unified model for predicting human Attention and Responses on visual content (NeurIPS 2024 · similarity 0.85)
- Probabilistic Prompt Adaptation for Unified Image Aesthetics and Quality Assessment (CVPR 2026 · similarity 0.85)
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.
