Research map / Reinforcement learning / Policy gradient and actor-critic
Discovering Reinforcement Learning Algorithms
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 |
|---|---|---|---|
| Discover prediction semantics as well as the policy update rule, rather than relying on a TD(λ) value function | LPG · Toy training environments | episode return | Helped |
| Train an update rule on toy environments and apply it to unseen Atari games | LPG · Atari (57 games) | games outperforming humans: 14 games | Helped |
| Include delayed chain MDPs in the meta-training environments (ablation) | LPG · Atari | generalisation performance | Helped |
| Use softmax categorical predictions and KL-divergence prediction updates instead of real-valued predictions and squared error (ablation) | LPG · Toy training environments | episode return | Helped |
| Add entropy regularization for predictions (ablation) | LPG · Toy training environments | episode return | Helped |
| Add L2 regularization on policy and prediction targets (ablation) | LPG · Toy training environments | episode return | Helped |
| Balance agent hyperparameters during meta-training rather than fixing them per environment (ablation) | LPG · Toy training environments | episode return | Helped |
| Subtract an action-independent baseline in the agent update (ablation) | LPG · Atari | episode return | Helped |
Closest papers by meaning
- Discovering Temporally-Aware Reinforcement Learning Algorithms (ICLR 2024 · similarity 0.86)
- Transformers Can Learn Temporal Difference Methods for In-Context Reinforcement Learning (ICLR 2025 · similarity 0.82)
- Can Learned Optimization Make Reinforcement Learning Less Difficult? (NeurIPS 2024 · similarity 0.82)
- First-Explore, then Exploit: Meta-Learning to Solve Hard Exploration-Exploitation Trade-Offs (NeurIPS 2024 · similarity 0.82)
- Discerning Temporal Difference Learning (AAAI 2024 · similarity 0.82)
- Meta-Reinforcement Learning with Universal Policy Adaptation: Provable Near-Optimality under All-task Optimum Comparator (NeurIPS 2024 · similarity 0.81)
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.