Research map / Reinforcement learning / Policy gradient and actor-critic
Curriculum-Based Asymmetric Multi-Task Reinforcement Learning
Hanchi Huang, Deheng Ye, Li Shen, Wei Liu
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 mode switching and multiple ranking functions (vs AMRL) (ablation) | CAMRL · MT10 ablation (avg of 10 tasks) | avg reward: 1844.9 -> 5638.7 | Improved |
| add mode switching (ablation) | CAMRL · MT10 ablation (avg of 10 tasks) | avg reward: 2779.5 -> 5638.7 | Improved |
| use finite a in indicator (vs a=positive infinity) (ablation) | CAMRL · MT10 ablation (avg of 10 tasks) | avg reward: 4571.3 -> 5638.7 | Improved |
| use finite b in indicator (vs b=positive infinity) (ablation) | CAMRL · MT10 ablation (avg of 10 tasks) | avg reward: 4465.1 -> 5638.7 | Improved |
| use nonzero c (vs c=0) (ablation) | CAMRL · MT10 ablation (avg of 10 tasks) | avg reward: 4929.1 -> 5638.7 | Improved |
| use nonzero d in tanh ranking (vs d=0) (ablation) | CAMRL · MT10 ablation (avg of 10 tasks) | avg reward: 3255.7 -> 5638.7 | Improved |
| use nonzero lambda2 (vs 0) (ablation) | CAMRL · MT10 ablation (avg of 10 tasks) | avg reward: 4525.4 -> 5638.7 | Improved |
| use nonzero lambda3 (vs 0) (ablation) | CAMRL · MT10 ablation (avg of 10 tasks) | avg reward: 5386.8 -> 5638.7 | Improved |
| use nonzero lambda4 (vs 0) (ablation) | CAMRL · MT10 ablation (avg of 10 tasks) | avg reward: 5528.7 -> 5638.7 | Improved |
| add prior-knowledge differentiable ranking loss | CAMRL · MT10 | reward: better on 5 of 10 tasks | Mixed |
| add prior-knowledge differentiable ranking loss | CAMRL · MT50 | reward: better on 38 of 50 tasks | Improved |
| add prior-knowledge differentiable ranking loss | CAMRL · Atari | reward: better on 8 of 10 tasks | Improved |
Closest papers by meaning
- AMAGO-2: Breaking the Multi-Task Barrier in Meta-Reinforcement Learning with Transformers (NeurIPS 2024 · similarity 0.86)
- SRL: Scaling Distributed Reinforcement Learning to Over Ten Thousand Cores (ICLR 2024 · similarity 0.86)
- AdaCuRL: Adaptive Curriculum Reinforcement Learning with Invalid Sample Mitigation and Historical Revisiting (AAAI 2026 · similarity 0.86)
- Dual RL: Unification and New Methods for Reinforcement and Imitation Learning (ICLR 2024 · similarity 0.86)
- Towards an Information Theoretic Framework of Context-Based Offline Meta-Reinforcement Learning (NeurIPS 2024 · similarity 0.85)
- MSRL: Scaling Generative Multimodal Reward Modeling via Multi-Stage Reinforcement Learning (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.
