Events

Machine Learning for a Changing Climate: Forecast, Risk, Adapt

Speaker: Cynthia Zeng

Location: 370 Jay Street, Room 1201

Date: Wednesday, April 9, 2025

Climate change is intensifying the frequency and severity of natural disasters across the globe, making societal adaptation an urgent priority. In this talk, I present two key avenues where machine learning can play a transformative role in addressing climate adaptation challenges. The first part introduces a multimodal machine learning framework designed for natural disaster prediction. This flexible framework integrates diverse data types—such as images, text, and tabular data—enabling predictions across both short-term and long-term timeframes. For instance, in 24-hour hurricane forecasting, we show that ML models can improve the National Hurricane Center's current forecasts. Accurate long-term ML-driven risk assessments have profound implications for urban planning, infrastructure investments, and insurance policies, helping to shape more resilient societies. In the second part, I explore how ML-predicted risks can be integrated into catastrophe insurance pricing through Robust Optimization. We demonstrate the effectiveness of this approach using data from the US National Flood Insurance Program (NFIP), showcasing its potential to improve pricing accuracy and risk management. This talk aims to open a broader conversation about the applications of machine learning in tackling global climate challenges, and how we can collaboratively leverage these technologies for a more sustainable future.