NeurIPS 2023 accepted papers by research area
3,584 papers were accepted at NeurIPS 2023 (+683 vs 2022). Here is how they split across 20 research areas, and which topics gained the most ground.
See NeurIPS 2023 in the interactive map
Research areas
| Area | Papers | Share | Change in share vs 2022 |
|---|---|---|---|
| Learning theory and algorithms | 426 | 11.9% | -2.1 pts |
| Probabilistic and scientific modeling | 380 | 10.6% | +0.2 pts |
| Label-efficient and robust learning | 342 | 9.5% | -2.4 pts |
| Architectures and efficiency | 275 | 7.7% | -1.5 pts |
| Reinforcement learning | 273 | 7.6% | -0.4 pts |
| Image generation | 253 | 7.1% | +4.1 pts |
| Large language models | 215 | 6.0% | +3.4 pts |
| Graph learning | 178 | 5.0% | steady |
| Optimization | 176 | 4.9% | -1.9 pts |
| Vision-language and multimodal | 161 | 4.5% | +1.8 pts |
| Bandits and online learning | 130 | 3.6% | -1.3 pts |
| Federated learning and privacy | 130 | 3.6% | +0.7 pts |
| 3D vision | 120 | 3.3% | +0.1 pts |
| Adversarial robustness and security | 117 | 3.3% | -0.4 pts |
| Detection and segmentation | 107 | 3.0% | steady |
| Multi-agent systems and games | 86 | 2.4% | -0.8 pts |
| Video | 85 | 2.4% | +0.4 pts |
| Time series | 52 | 1.5% | +0.6 pts |
| Point clouds | 46 | 1.3% | steady |
| Continual learning | 32 | 0.9% | -0.3 pts |
Fastest-growing topics at NeurIPS 2023
Change in share of NeurIPS papers from 2022 to 2023, topics with at least 10 papers.
- Fast diffusion sampling (Image generation): 16 → 87 papers
- Text-to-image diffusion (Image generation): 7 → 52 papers
- Molecules and proteins (Probabilistic and scientific modeling): 43 → 93 papers
- Human and embodied modeling (Vision-language and multimodal): 30 → 74 papers
- Language, safety and interpretability (Large language models): 39 → 82 papers
- Autoregressive and latent generation (Image generation): 29 → 60 papers
- Agents, code and math (Large language models): 15 → 39 papers
- Instruction and fine-tuning (Large language models): 0 → 17 papers
- Reasoning and chain of thought (Large language models): 2 → 19 papers
- Open-set and few-shot detection (Detection and segmentation): 37 → 60 papers