Research map / Learning theory and algorithms / Prediction and decision losses
A Batch Learning Framework for Scalable Personalized Ranking
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 batch rank-sensitive training instead of CE classification loss | BARS · Yelp | NDCG: 2.6 -> 3.0 | Helped |
| use batch rank-sensitive training instead of WARP pairwise rank estimation | BARS · ML-20m | Precision: 10.1 -> 10.2 | Helped |
| use batch rank-sensitive training instead of WARP pairwise rank estimation | BARS · XING | NDCG: 6.7 -> 7.1 | Helped |
| add margin formulation to smoothed rank approximation (ablation) | BARS · ML-20m | NDCG: 13.6 -> 13.9 | Helped |
| add margin formulation to smoothed rank approximation (ablation) | BARS · Yelp | NDCG: 2.9 -> 3.0 | Helped |
| add margin formulation to smoothed rank approximation (ablation) | BARS · XING | NDCG: 6.9 -> 6.9 | No clear effect |
| use polynomial rather than logarithmic rank-sensitive loss (ablation) | BARS with margin rank · ML-20m | NDCG: 13.9 -> 13.9 | No clear effect |
| use polynomial rather than logarithmic rank-sensitive loss (ablation) | BARS with margin rank · Yelp | NDCG: 2.9 -> 2.9 | No clear effect |
| use polynomial rather than logarithmic rank-sensitive loss (ablation) | BARS with margin rank · XING | NDCG: 6.9 -> 7.1 | No clear effect |
| suppress large margin violations in the rank estimate (ablation) | BARS with logarithmic loss · Yelp | NDCG: 2.9 -> 3.0 | Helped |
| reduce item sample rate from 1.0 to 0.1 (ablation) | BARS · ML-20m development | NDCG: 16.0 -> 15.9 | No clear effect |
| reduce item sample rate from 1.0 to 0.05 (ablation) | BARS · Yelp development | NDCG: 3.0 -> 2.9 | No clear effect |
Closest papers by meaning
- Extreme Learning to Rank via Low Rank Assumption (ICML 2018 · similarity 0.87)
- From Pairwise to Ranking: Climbing the Ladder to Ideal Collaborative Filtering with Pseudo-Ranking (AAAI 2025 · similarity 0.87)
- SQL-Rank: A Listwise Approach to Collaborative Ranking (ICML 2018 · similarity 0.87)
- Adaptively Learning to Select-Rank in Online Platforms (ICML 2024 · similarity 0.84)
- Preference Diffusion for Recommendation (ICLR 2025 · similarity 0.84)
- Recommendations with Sparse Comparison Data: Provably Fast Convergence for Nonconvex Matrix Factorization (ICML 2025 · similarity 0.84)
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.