Research map / Reinforcement learning / Visual and representation RL
Curriculum and environment design
Generates training environments and curricula to improve agent generalization and skill acquisition.
81 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Visual and representation RL, keyword cluster “search · decision · systems”
State History Tried Checklist Preprints Who Export
Where it stands
Not enough benchmark data Fewer than four papers report the same benchmark, so there is no record to follow.
21 papers in 2023–24 and 14 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 3.7% to 2.0%.
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2016 · Adaptive Skills Adaptive Partitions (ASAP) (NeurIPS 2016 · 39 citations)
- 2017 · Reinforcement Learning with Unsupervised Auxiliary Tasks (ICLR 2017 · 259 citations)
- 2018 · Continuous Adaptation via Meta-Learning in Nonstationary and Competitive Environments (ICLR 2018 · 150 citations)
- 2019 · Recurrent Experience Replay in Distributed Reinforcement Learning (ICLR 2019 · 252 citations)
- 2020 · Automatic Curriculum Learning through Value Disagreement (NeurIPS 2020 · 39 citations)
- 2021 · A Policy Gradient Algorithm for Learning to Learn in Multiagent Reinforcement Learning (ICML 2021 · 12 citations)
- 2022 · Evolving Curricula with Regret-Based Environment Design (ICML 2022 · 19 citations)
- 2023 · Information Design in Multi-Agent Reinforcement Learning (NeurIPS 2023 · 5 citations)
- 2024 · Sustainability of Data Center Digital Twins with Reinforcement Learning (AAAI 2024 · 7 citations)
- 2025 · GenAL: Generative Agent for Adaptive Learning (AAAI 2025 · 9 citations)
- 2026 · Adaptive Reinforcement Learning for Unobservable Random Delays (ICML 2026 · 0 citations)
Earliest papers here
- Adaptive Skills Adaptive Partitions (ASAP) (NeurIPS 2016 · 39 citations)
- Reuse of Neural Modules for General Video Game Playing (AAAI 2016 · 20 citations)
- Reinforcement Learning with Unsupervised Auxiliary Tasks (ICLR 2017 · 259 citations)
What has been tried
69 papers here report 676 measured changes (a component added, removed or swapped, compared with a baseline): helped 62%, no clear effect 12%, hurt 22%, mixed 4%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Remove textual item content and recommend using item IDs only | Junyi, ASSIST09, TextLog · Ep | Hurt | GenAL: Generative Agent for Adaptive Learning (AAAI 2025 · 9 citations) | |
| add the AIM reward rφ to curriculum goal generation | PointSpiralMaze · success rate | Helped | Diffusion-based Curriculum Reinforcement Learning (NeurIPS 2024 · 3 citations) | |
| Add top-level geographic workload distribution | Data center cluster, load shifting · CO2 emissions (tons) | 3845 ± 3 -> 3515 ± 3 | Helped | Hierarchical Multi-Agent Framework for Carbon-Efficient Liquid-Cooled Data Center Clusters (AAAI 2025 · 2 citations) |
| Use 50% negative experiences in training batches | Training experiments over 5000 steps · performance | 50% | Helped | Of Mice and Machines: A Comparison of Learning Between Real World Mice and RL Agents (ICML 2025 · 0 citations) |
| disable option execution in the environment and use options only for MCTS planning | 26 Atari games · mean human-normalized score | 1054.30 % -> 1008.15 % | Hurt | OptionZero: Planning with Learned Options (ICLR 2025 · 0 citations) |
| Add timeout-nonterminal bootstrapping | EARL benchmark · performance | Helped | Intelligent Switching for Reset-Free RL (ICLR 2024 · 0 citations) | |
| remove inter-episode opponent modeling | MSH-4h2s self-play · normalized score | 0.9767 -> 0.9708 | No clear effect | Efficient Adaptation in Mixed-Motive Environments via Hierarchical Opponent Modeling and Planning (ICML 2024 · 0 citations) |
| Train on multi-trajectory sequences instead of single-trajectory sequences | MiniHack-Labyrinth · episodic return | -0.323 ± 0.05 -> 0.780 ± 0.07 | Helped | Generalization to New Sequential Decision Making Tasks with In-Context Learning (ICML 2024 · 0 citations) |
| Use novelty alone for replay prioritization | Minigrid held-out environments · solved rate | Helped | Improving Environment Novelty Quantification for Effective Unsupervised Environment Design (NeurIPS 2024 · 0 citations) | |
| Add sample-mask updates during stage 1 | Alchemy · target-region archive coverage | Helped | Enabling Adaptive Agent Training in Open-Ended Simulators by Targeting Diversity (NeurIPS 2024 · 0 citations) | |
| Mask out selected training contexts | Coinrun, Heist, Ninja, Climber, and other Procgen games · final average score | Mixed | Implicit Curriculum in Procgen Made Explicit (NeurIPS 2024 · 0 citations) |
What reviewers will expect
Datasets and baselines named in the 11 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Baselines
- DQN · 2 of 11 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 35 papers; from OpenAlex and the papers' affiliations).
- Singapore Management University · 3 papers
- University of Southern California · 3 papers
- Hewlett Packard Enterprise · 2 papers
- Tsinghua University · 2 papers
- University College London · 2 papers
- University of Texas, Austin · 2 papers
- University of Texas at Austin · 2 papers
Use it
Download BibTeX of 40 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- In-context and adaptive RL (202 papers)
- Scalable RL systems and environments (179 papers)
- Probabilistic and autonomous RL (177 papers)
- RL for combinatorial routing (158 papers)
- RL with tree search for discovery (121 papers)
- Inferring decision-making from behavior (93 papers)
Elsewhere in the topic
- Learned state and action representations (148 papers)
- Dexterous manipulation transfer (141 papers)
- Skill learning and diffusion policies (86 papers)
- Robust visual encoders for RL (86 papers)
- Novelty and impact-driven exploration (85 papers)
- Hierarchical skills and priors (84 papers)
- Latent-space world models (83 papers)
- Goal-conditioned policy learning (66 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).