Research map / Reinforcement learning / Policy gradient and actor-critic
Reinforcement Learning with a Disentangled Universal Value Function for Item Recommendation
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 GoalRec instead of the strongest reward baseline | GoalRec · MovieLens, Taobao, and YooChoose vanilla simulations | cumulative reward: averaged increase of 3.7%, 6.1%, 10.7%, respectively | Helped |
| Use GoalRec instead of Rainbow | GoalRec · MovieLens high-dimensional simulation | cumulative reward: 28.2 -> 33.8 | Helped |
| Use GoalRec instead of Rainbow | GoalRec · Taobao high-dimensional simulation | cumulative reward: 2.61 -> 3.24 | Helped |
| Use GoalRec instead of Rainbow | GoalRec · YooChoose high-dimensional simulation | cumulative reward: 3.60 -> 4.35 | Helped |
| Use GoalRec instead of strongest reward baseline | GoalRec · MovieLens, Taobao, and YooChoose high-variance simulations | cumulative reward: 32.1 vs 29.5; 3.02 vs 2.82; 4.05 vs 3.92 | Helped |
| Train GoalRec from offline trajectories rather than simulator interaction | GoalRec-off · vanilla simulations | cumulative reward: 29.1 vs 31.6; 3.20 vs 3.14; 4.09 vs 4.23 | Mixed |
| Optimize long-term trajectory reward rather than item-level click prediction | GoalRec · vanilla simulations | CTR: MovieLens: 0.601 vs 0.544; Taobao: 0.024 vs 0.021; YooChoose: 0.041 vs 0.037 | Hurt |
Closest papers by meaning
- On the Unexpected Effectiveness of Reinforcement Learning for Sequential Recommendation (ICML 2024 · similarity 0.89)
- UNEX-RL: Reinforcing Long-Term Rewards in Multi-Stage Recommender Systems with UNidirectional EXecution (AAAI 2024 · similarity 0.88)
- The Adaptive Q-Network for Recommendation Tasks with Dynamic Item Space (AAAI 2025 · similarity 0.88)
- Objective Shaping with Hard Negatives: Windowed Partial AUC Optimization for RL-based LLM Recommenders (NeurIPS 2026 · similarity 0.87)
- Think before Recommendation: Autonomous Reasoning-enhanced Recommender (NeurIPS 2025 · similarity 0.87)
- R$^2$ec: Towards Large Recommender Models with Reasoning (NeurIPS 2025 · similarity 0.87)
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.