Research map / Reinforcement learning / Offline RL
Towards Off-Policy Learning for Ranking Policies with Logged Feedback
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 sequential value ranking (VR) instead of off-policy policy gradient with SASRec | SASRec · YooChoose | HR@5: 0.2968 -> 0.3187 | Helped |
| Use sequential value ranking (VR) instead of off-policy policy gradient with SASRec | SASRec · RetailRocket | NDCG@5: 0.1638 -> 0.1729 | Helped |
| Use VR rather than MLE | SASRec, GRU4Rec, and Caser · YooChoose and RetailRocket | ranking metrics HR@5, NDCG@5, HR@20, and NDCG@20 | Helped |
| Add importance weighting to the TD loss (ablation) | VR · YooChoose | NDCG@5: 19.21±0.6 -> 19.92± 0.9; 19.52±0.7 -> 20.43±1.1; 18.02±0.5 -> 19.56±1.5; 17.67±0.6 -> 18.92±1.8 | Helped |
| Increase the contribution of future rewards with the discount factor γ (ablation) | VR · YooChoose and RetailRocket | NDCG@5 | Mixed |
| Assign higher reward to purchases than clicks (ablation) | VR · YooChoose and RetailRocket multi-objective setting | purchase and click performance | Helped |
| Use VR instead of the behavior policy | RecSim recommendation policy · RecSim online evaluation | CTR | Helped |
Closest papers by meaning
- Unified Off-Policy Learning to Rank: a Reinforcement Learning Perspective (NeurIPS 2023 · similarity 0.91)
- A General Framework for Off-Policy Learning with Partially-Observed Reward (ICLR 2025 · similarity 0.86)
- Probabilistic Offline Policy Ranking with Approximate Bayesian Computation (AAAI 2024 · similarity 0.86)
- Objective Shaping with Hard Negatives: Windowed Partial AUC Optimization for RL-based LLM Recommenders (NeurIPS 2026 · similarity 0.86)
- Preference-based Reinforcement Learning beyond Pairwise Comparisons: Benefits of Multiple Options (NeurIPS 2025 · similarity 0.86)
- Distributional Off-Policy Evaluation for Slate Recommendations (AAAI 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.