Robust Calibration For Improved Weather Prediction Under Distributional Shift
This addresses robust weather forecasting for meteorologists under real-world distribution shifts, but appears incremental as it builds on existing techniques.
The paper tackles improving weather prediction accuracy and uncertainty calibration under distributional shift by combining mixture of experts with computer vision data augmentation and post-hoc calibration, achieving potentially better results than boosted tree models for tabular data.
In this paper, we present results on improving out-of-domain weather prediction and uncertainty estimation as part of the \texttt{Shifts Challenge on Robustness and Uncertainty under Real-World Distributional Shift} challenge. We find that by leveraging a mixture of experts in conjunction with an advanced data augmentation technique borrowed from the computer vision domain, in conjunction with robust \textit{post-hoc} calibration of predictive uncertainties, we can potentially achieve more accurate and better-calibrated results with deep neural networks than with boosted tree models for tabular data. We quantify our predictions using several metrics and propose several future lines of inquiry and experimentation to boost performance.