ICLR 2022 accepted papers by research area
1,095 papers were accepted at ICLR 2022 (+235 vs 2021). Here is how they split across 20 research areas, and which topics gained the most ground.
See ICLR 2022 in the interactive map
Submitting to ICLR? Checklist →
Research areas
| Area | Papers | Share | Change in share vs 2021 |
|---|---|---|---|
| Label-efficient and robust learning | 152 | 13.9% | -2.3 pts |
| Architectures and efficiency | 138 | 12.6% | -2.0 pts |
| Reinforcement learning | 125 | 11.4% | +0.5 pts |
| Probabilistic and scientific modeling | 109 | 10.0% | -0.3 pts |
| Learning theory and algorithms | 93 | 8.5% | -1.2 pts |
| Graph learning | 81 | 7.4% | +3.0 pts |
| Optimization | 65 | 5.9% | -0.1 pts |
| Adversarial robustness and security | 54 | 4.9% | steady |
| Image generation | 46 | 4.2% | -0.5 pts |
| Detection and segmentation | 35 | 3.2% | +0.6 pts |
| Multi-agent systems and games | 33 | 3.0% | +0.1 pts |
| Federated learning and privacy | 27 | 2.5% | +0.5 pts |
| Large language models | 25 | 2.3% | +0.5 pts |
| Time series | 25 | 2.3% | +0.8 pts |
| Continual learning | 24 | 2.2% | +0.3 pts |
| Vision-language and multimodal | 21 | 1.9% | +0.4 pts |
| 3D vision | 17 | 1.6% | +0.4 pts |
| Bandits and online learning | 10 | 0.9% | +0.6 pts |
| Video | 8 | 0.7% | -1.4 pts |
| Point clouds | 7 | 0.6% | steady |
Fastest-growing topics at ICLR 2022
Change in share of ICLR papers from 2021 to 2022, topics with at least 10 papers.
- Graph self-supervised learning (Graph learning): 2 → 20 papers
- Attention and transformers (Architectures and efficiency): 17 → 35 papers
- Visual and representation RL (Reinforcement learning): 41 → 65 papers
- Domain adaptation and generalization (Label-efficient and robust learning): 5 → 19 papers
- GNN architectures (Graph learning): 26 → 43 papers
- Human and embodied modeling (Vision-language and multimodal): 2 → 10 papers
- Graph algorithms and clustering (Graph learning): 2 → 10 papers
- Nonconvex and smooth optimization (Optimization): 8 → 16 papers
- Language, safety and interpretability (Large language models): 9 → 17 papers
- Open-set and few-shot detection (Detection and segmentation): 6 → 13 papers
More venues
Same year: ICML 2022 · NeurIPS 2022 · CVPR 2022 · AAAI 2022 · ECCV 2022 · TPAMI 2022 · IJCV 2022 · JMLR 2022
ICLR in other years: 2017 · 2018 · 2019 · 2020 · 2021 · 2023 · 2024 · 2025 · 2026
