arXiv：cs.LG· Jannik Wiese, Johannes Schusterbauer, Tommaso Martorella, Bj\"orn Ommer·· 4 小时前
Just Weather Scoring：用分布扩散实现高效端到端临近降水预报
Just Weather Scoring: Efficient End-to-end Nowcasting with Distributional Diffusion
AI 导读
JWS 是一种单阶段端到端扩散模型，直接在雷达空间预报并支持少步生成，无需单独训练的压缩或确定性预报组件。它结合 Masked Asynchronous Diffusion 与 scoring-rule 目标，在 SEVIR 和 MeteoNet 上取得 SOTA 概率预报性能，同时降低训练与推理成本。最小模型以更少参数保持竞争力，推理速度提升超过 17 倍。
正文
Abstract:Generative diffusion models are well-suited for probabilistic precipitation nowcasting, but existing approaches often rely on separately trained compression or deterministic forecasting components and remain costly at inference due to iterative denoising. We introduce Just Weather Scoring (JWS), a single-stage, end-to-end diffusion model which addresses both issues by forecasting directly in radar space and enabling few-step generation. Radar-space modeling greatly simplifies training and inference and eliminates uncertainty arising from lossy compression. JWS combines Masked Asynchronous Diffusion, a timestep-sampling scheme that preserves clean context while adapting diffusion training to high-dimensional spatio-temporal data, with a simple scoring-rule objective that aligns training with probabilistic forecasting and unlocks few-step generation. On the SEVIR and MeteoNet benchmarks, JWS achieves state-of-the-art probabilistic forecasting performance at reduced training and inference cost. Even our smallest model remains competitive using substantially fewer parameters and more than 17x faster inference.
| Comments: | Project Page: this https URL |
| Subjects: | Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV) |
| Cite as: | arXiv:2610.12189 [cs.LG] |
| (or arXiv:2610.12189v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.12189 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Johannes Schusterbauer [view email]
[v1]
Thu, 8 Oct 2026 15:50:43 UTC (21,624 KB)
来源：arXiv：cs.LG · arxiv.org