arXiv cs.AI· Joseph Rawson, Domniki Ladopoulou, Petros Dellaportas·· 2 天前AI 评分20
多任务神经扩散过程（MT-NDPs）：面向不确定性感知多任务回归与少样本适应
Multi-task Neural Diffusion Processes
AI 导读
研究者提出多任务神经扩散过程（MT-NDPs），通过引入任务编码器从上下文观测中提取低维表示，使扩散模型具备任务条件化的概率回归与少样本适应能力，并保持对上下文观测顺序的置换不变性。
正文 · 原文
Abstract:Neural diffusion processes provide a scalable, non-Gaussian approach to modelling distributions over functions, but existing formulations are limited to single-task inference and do not capture dependencies across related tasks. We propose multi-task neural diffusion processes (MT-NDPs), an extension that incorporates a task encoder to enable task-conditioned probabilistic regression and few-shot adaptation across related functions. The task encoder extracts a low-dimensional representation from context observations and conditions the diffusion model on this representation, allowing information sharing across tasks while accommodating variable-sized context sets and preserving permutation invariance with respect to the ordering of context observations. We evaluate MT-NDPs through synthetic experiments and two real-world applications involving wind turbine data and school examination outcomes. The results demonstrate improved predictive accuracy relative to single-task diffusion baselines when context observations are available, alongside competitive performance against neural-process and Gaussian-process alternatives. In the wind-energy application, MT-NDPs achieve the lowest average coverage error among the evaluated models at larger context sizes, demonstrating improved predictive uncertainty calibration. Overall, our findings highlight the potential of task-conditioned diffusion models for uncertainty-aware multi-task regression and few-shot adaptation to previously unseen tasks.
| Comments: | 37 pages, 16 figures, 5 tables |
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Applications (stat.AP); Machine Learning (stat.ML) |
| Cite as: | arXiv:2510.03419 [cs.LG] |
| (or arXiv:2510.03419v3 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2510.03419 arXiv-issued DOI via DataCite |
Submission history
From: Domniki Ladopoulou [view email]
[v1]
Fri, 3 Oct 2025 18:26:23 UTC (25,395 KB)
[v2]
Fri, 16 Jan 2026 12:40:10 UTC (25,414 KB)
[v3]
Thu, 15 Oct 2026 12:14:37 UTC (27,631 KB)
来源：arXiv cs.AI · arxiv.org