The paper “Fine-tuning for Extreme Event Prediction: Are Ensemble Methods All You Need?“, co-authored by Imran Nasim (IBM UK and Maths@Surrey) and Joao Lucas de Sousa Almeida (IBM Research, Brazil), was accepted for presentation at KDD 2025, the world’s premier conference on Knowledge Discovery and Data Mining. It was held this year in Toronto, Canada, from 3-7 August (Conference link here). The paper was accepted for oral presentation and Imran presented it virtually. AI-driven weather forecasting models have achieved significant advancements in both speed and accuracy. However, accurately forecasting rare, high-impact extreme events, such as storms and heatwaves, remains a critical challenge. In this study, the authors investigate uncertainty-aware extreme event forecasting using the recently introduced time-series foundation model, Tiny Time Mixers (TTM). A link to the full 17 page paper is here. The image below shows Figure 7 from the paper. Click on image to enlarge.