Generating scenarios for extreme events, without extreme data
A new algorithm learns to anticipate the unprecedented scenarios that critical infrastructure and global supply chains are least prepared for.
Researchers at MIT have developed a machine-learning algorithm that can generate plausible, extreme events and worst-case scenarios, without relying on historical data of such events. This method, dubbed "Extreme Event Aware" or "η-learning," aims to quantify the characteristics of unprecedented extreme events, like Hurricane Katrina or a once-in-a-century rainfall event.
By learning statistical relationships between point statistics (e.g., maximum rainfall) and spatial maps (e.g., precipitation patterns), the algorithm can create plausible scenarios that are riskier than any previously observed events, yet still plausible. This tool can help city planners, policymakers, and insurance companies estimate the risk of extreme weather events, such as determining the impact of a once-every-100-year storm on New York City.
Written by urgent.news from MIT News Research's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
This story
This is one outlet's version. Read the fullest account.
