Research map / Reinforcement learning
Imitation learning: research map
450 papers on Imitation learning in Reinforcement learning, from ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR (2016–2026), grouped into 2 clusters and 6 approaches. The busiest year so far is 2026.
What it is
Instead of a reward, the agent gets demonstrations of the task and learns to imitate them, either by copying actions directly (behaviour cloning) or by inferring the reward the expert was optimising (inverse RL). Robotics and driving are the main uses.
Terms you will see on this page:
- Behaviour cloning
- supervised learning from expert state-action pairs.
- Inverse RL (IRL)
- recovering the reward that explains expert behaviour.
- Adversarial imitation
- training a policy so a discriminator cannot tell it from the expert.
Start reading with the most cited papers here:
- Cooperative Inverse Reinforcement Learning (NeurIPS 2016 · 387 citations)
- Generative Adversarial Imitation Learning (NeurIPS 2016 · 353 citations)
Within Reinforcement learning, its share shrank from 7.6% in 2023–24 to 5.0% in 2025–26 (119 → 116 papers at ICML, NeurIPS, ICLR, CVPR and AAAI, the venues with data for all four years).
Explore Imitation learning in the interactive map
Working on something in this topic? Describe your idea in scime atlas to see which approach it falls under, the closest papers by meaning and how crowded the spot has become.
What changed recently
Approaches whose share of the topic moved most between 2023–24 and 2025–26 (ICML, NeurIPS, ICLR, CVPR and AAAI):
- Imitation combined with RL rising: 22% → 33% of the topic (26 → 38 papers)
- Offline imitation from imperfect data rising: 22% → 27% of the topic (26 → 31 papers)
- Theory of inverse RL falling: 15% → 7% of the topic (18 → 8 papers)
- Behavior cloning theory and models falling: 19% → 15% of the topic (23 → 17 papers)
Papers from 2025–2026 that are already cited most:
- Inverse Reinforcement Learning by Estimating Expertise of Demonstrators (AAAI 2025 · 3 citations)
Approaches and key papers
Imitation from demonstrations · 333 papers
Keywords: il, adversarial imitation, cloning
Approaches in this cluster:
- Offline imitation from imperfect data (109 papers)
Imitate from suboptimal or limited demonstrations using distribution correction and offline data. - Imitation combined with RL (112 papers)
Mix demonstrations with reinforcement learning, including hierarchical and self-correcting policies. - Behavior cloning theory and models (73 papers)
Analyze behavior cloning and build flexible imitative or foundation-model-based policies. - Adversarial imitation learning (39 papers)
Match expert behavior with GAN-style discriminators and provably efficient adversarial objectives.
Most cited and most cited since 2024:
- Generative Adversarial Imitation Learning (NeurIPS 2016 · 353 citations)
- Zero-Shot Visual Imitation (ICLR 2018 · 196 citations)
- BAKU: An Efficient Transformer for Multi-Task Policy Learning (NeurIPS 2024 · 17 citations)
- No Prior Mask: Eliminate Redundant Action for Deep Reinforcement Learning (AAAI 2024 · 13 citations)
Inverse reinforcement learning · 117 papers
Keywords: inverse reinforcement, irl, reward function
Approaches in this cluster:
- Reward learning from demonstrations (72 papers)
Infer reward functions from suboptimal or limited demonstrations with scalable inverse RL algorithms. - Theory of inverse RL (45 papers)
Give theoretical guarantees and exploration strategies for offline and active inverse reinforcement learning.
Most cited and most cited since 2024:
- Cooperative Inverse Reinforcement Learning (NeurIPS 2016 · 387 citations)
- Predicting Goal-Directed Human Attention Using Inverse Reinforcement Learning (CVPR 2020 · 100 citations)
- Meta-Inverse Reinforcement Learning for Mean Field Games via Probabilistic Context Variables (AAAI 2024 · 5 citations)
- In-Trajectory Inverse Reinforcement Learning: Learn Incrementally Before an Ongoing Trajectory Terminates (NeurIPS 2024 · 3 citations)
Related topics in Reinforcement learning
- MDPs and planning (726)
- Visual and representation RL (2,264)
- Policy gradient and actor-critic (1,570)
- RL from feedback (758)
- Offline RL (646)
