Scalable Uncertainty Quantification for Extreme Weather Forecasting via Empirical Neural Tangent Kernels

📰 ArXiv cs.AI

Learn to quantify uncertainty in extreme weather forecasting using Neural Tangent Kernels for better decision-making

advanced Published 3 Jun 2026

Action Steps

Apply Neural Tangent Kernel-based uncertainty quantification to existing deep learning weather models
Use last-layer empirical features to estimate uncertainty
Evaluate the quality of uncertainty quantification using theoretical analysis
Compare the performance of different architectures for uncertainty quantification
Implement NTK-UQ in a real-world weather forecasting system to test its scalability

Who Needs to Know This

Data scientists and researchers working on weather forecasting models can benefit from this approach to improve the accuracy and reliability of their predictions

Key Insight

💡 Neural Tangent Kernel-based uncertainty quantification can provide accurate and reliable uncertainty estimates for deep learning weather models

Key Takeaways

Learn to quantify uncertainty in extreme weather forecasting using Neural Tangent Kernels for better decision-making

Full Article

Title: Scalable Uncertainty Quantification for Extreme Weather Forecasting via Empirical Neural Tangent Kernels

Abstract:
arXiv:2606.02886v1 Announce Type: cross Abstract: Deep learning weather models now match numerical weather prediction accuracy while running orders of magnitude faster, but produce deterministic forecasts without uncertainty estimates, a critical gap for high-stakes decisions during extreme weather events. This paper proposes Neural Tangent Kernel-based uncertainty quantification (NTK-UQ) using last-layer empirical features. Theoretical analysis predicts that UQ quality is architecture-dependent

Read full paper → ← Back to Reads