发表机构
School of Computer Science and Engineering, The Hebrew University of Jerusalem; Institute of Earth Sciences, The Hebrew University of Jerusalem; Hebrew University Business School, The Hebrew University of Jerusalem（耶路撒冷希伯来大学计算机科学与工程学院; 耶路撒冷希伯来大学地球科学研究所; 耶路撒冷希伯来大学商学院）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对深度学习模型‘黑箱’问题阻碍洪水预测信任度的挑战，提出上下文感知概念蒸馏框架CACD，引入无监督管道和残差超网络，经全球多流域评估，该模型高保真且优于黑箱基线，平衡了AI准确性与决策透明度。
AI 中文摘要
有效的洪水风险管理依赖于准确的预测，但最先进的深度学习模型的‘黑箱’性质给高风险公共安全决策中的信任和问责制带来了障碍。现有可解释人工智能（XAI）方法只能提供局部归因，无法提供灾害应对当局所需的可验证、具有操作意义的因果叙述。为应对这一社会挑战，我们提出了上下文感知概念蒸馏（CACD）框架，与领域专家合作开发，将不透明的长短期记忆网络（LSTM）提炼为可解释、具有水文意识的替代模型。我们引入了一个无监督管道来发现‘水文语言’和一个基于静态流域特征动态调制这些概念的残差超网络。在全球5203个流域进行评估，我们的模型实现了高保真度（中位数纳什效率系数为0.70），在未见的未来数据上显著优于黑箱基线（如多层感知器）。通过证明人类可解释的概念足以重建洪水动态，这项工作在人工智能准确性与负责任的环境决策所需的透明度之间取得了平衡。
英文摘要
Effective flood risk management relies on accurate forecasting, yet the "black box" nature of stateof-the-art Deep Learning models creates a barrier to trust and accountability in high-stakes public safety decisions. While existing Explainable AI (XAI) methods offer local attributions, they fail to provide the verifiable, operationally meaningful causal narratives required by disaster response authorities. To address this societal challenge, we propose Context-Aware Concept Distillation (CACD), a framework developed in collaboration with domain experts to distill opaque LSTMs into interpretable, hydrology-aware surrogate models. We introduce an unsupervised pipeline to discover a "Hydrological Language" and a Residual Hypernetwork that dynamically modulates these concepts based on static basin characteristics. Evaluated on 5,203 basins globally, our model achieves high fidelity (Median NSE 0.70), significantly outperforming black-box baselines (e.g., Multi Layer Perceptrons) on unseen future data. By demonstrating that human-interpretable concepts are sufficient to reconstruct flood dynamics, this work balances AI accuracy with the transparency required for responsible environmental decision-making.
Commentsto be published in IJCAI 2026 proceedings
