PapersSelected

用稀疏自编码器审计 GraphCast 内部机制并改进热带气旋预报

Auditing Learned Dynamics in Weather Models

Why we picked it

DeepMind 用稀疏自编码器拆解 GraphCast 的内部表征,还真能动手改:放大对流特征后,七个风暴的强度误差中位数降了 51%。做气象 AI 或模型可解释性的人都该看看。

DeepMind 在 arXiv 发表的论文以 GraphCast 为对象,用稀疏自编码器(SAE)提取特征,检验这些内部表征是否具有因果相关性。研究对 GraphCast 自回归 rollout 过程中的内部激活做干预,观察对预报结果的影响。

Body

实验有两个关键发现:一是消融一个与 GraphCast 计算网格绑定的特征,会使全球位势高度误差上升,但该特征并不对应传统物理语义;二是抑制对流和低层涡度相关特征会削弱预测的热带气旋加深(中心气压下降),而放大这些特征可降低强度误差,其中通过对流特征缩放在七个风暴上取得中位数 51% 的误差降低。

这项工作把可解释的物理关联与因果效用区分开来,说明高可信度场景下的 AI 预报系统审计不能只依赖语义标签,稀疏特征可以作为干预和学习化天气动力学的实验坐标。

Original · arXiv: Google DeepMind

Auditing Learned Dynamics in Weather Models

Learned weather models are experimentally accessible computational systems whose internal representations can be inspected and manipulated. Using sparse autoencoder features from GraphCast, DeepMind's autoregressive weather forecasting model, we test whether causal relevance requires physical semantics, a question that matters for auditing whether AI forecasters used in high-assurance settings rely on reliable mechanisms. We also look at whether physically associated latent directions can be steered to improve extreme-event forecasts, where interventions are applied to the model's internal activations during the rollout. We find that ablating a feature locked to GraphCast's computational mesh increases global geopotential height error despite lacking conventional physical semantics. Second, suppressing convection- and low-level-vorticity-associated features weakens predicted tropical-cyclone deepening (the fall in central pressure during intensification), while amplifying them reduces intensity error, with amplification of convection features through feature scaling achieving a median of 51\% error reduction over seven storms. Together, these results separate interpretable, physical association from causal utility and establish sparse features as experimental coordinates for auditing and intervening on learned weather dynamics.

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Papers" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include GraphCast, DeepMind, sparse autoencoders, interpretability.

Primary source verification: published directly by arXiv: Google DeepMind with first-party authority.