Research map / Reinforcement learning / Visual and representation RL
Collective Noise Contrastive Estimation for Policy Transfer Learning
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 |
|---|---|---|---|
| add playlist-data transfer via joint training (collective NCE) to radio-only NCE (ablation) | NCE-Radio · Xbox Music radio test data | log-likelihood per transition: -5.5197 -> -5.5072 | Helped |
| add playlist-data transfer via joint training (collective NCE) to radio-only NCE (ablation) | NCE-Radio · Xbox Music radio test data | IPS policy value: 0.3912 -> 0.4111 | Helped |
| use playlist-only trained model directly on radio data (no radio training) (ablation) | NCE · Xbox Music radio test data | log-likelihood per transition: -10.3978 vs -7.7932 for Random | Hurt |
| soft inter-domain regularization lambda2 versus none (ablation) | NCE-Collective · Xbox Music radio test data | IPS policy value | Helped |
| add L2 regularization lambda1=0.05 (ablation) | NCE-Collective and NCE-Radio · Xbox Music radio test data | log-likelihood convergence | Helped |
| increase noise samples k beyond 100 (ablation) | NCE-Collective · Xbox Music radio test data | log-likelihood per transition | Hurt |
Closest papers by meaning
- MultiScale Contextual Bandits for Long Term Objectives (NeurIPS 2025 · similarity 0.84)
- Efficient Policy Evaluation with Offline Data Informed Behavior Policy Design (ICML 2024 · similarity 0.84)
- A Snapshot of Influence: A Local Data Attribution Framework for Online Reinforcement Learning (NeurIPS 2025 · similarity 0.83)
- Contrastive Representation for Data Filtering in Cross-Domain Offline Reinforcement Learning (ICML 2024 · similarity 0.83)
- RL in Latent MDPs is Tractable: Online Guarantees via Off-Policy Evaluation (NeurIPS 2024 · similarity 0.83)
- Offline Data Enhanced On-Policy Policy Gradient with Provable Guarantees (ICLR 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.