Generalized Extreme Value

Generalized Extreme Value (GEV) distributions are used to model the statistical behavior of extreme events, such as floods or wildfires, focusing on the upper tail of the data distribution. Current research emphasizes developing computationally efficient methods for parameter estimation, often employing neural networks to overcome limitations of traditional maximum likelihood approaches and to handle complex relationships between predictors and extreme values. These advancements improve the accuracy and speed of risk assessment in various fields, enabling better prediction and management of extreme events with significant societal and environmental impacts.

Papers

May 7, 2023

Fast parameter estimation of Generalized Extreme Value distribution using Neural Networks
Sweta Rai, Alexis Hoffman, Soumendra Lahiri, Douglas W. Nychka, Stephan R. Sain, Soutir Bandyopadhyay
Neural Network Maximum Likelihood Parameter Estimation DIstribution Estimation Generalized Extreme Value

August 16, 2022

Regression modelling of spatiotemporal extreme U.S. wildfires via partially-interpretable neural networks
Jordan Richards, Raphaël Huser
Quantile Regression Regression Model Interpretable Neural Network Extreme Value Theory Regression Framework Generalized Extreme Value

May 5, 2022

DeepExtrema: A Deep Learning Approach for Forecasting Block Maxima in Time Series Data
Asadullah Hill Galib, Andrew McDonald, Tyler Wilson, Lifeng Luo, Pang-Ning Tan
Deep Neural Network Deep Learning Approach Time Series Data Extreme Value Generalized Extreme Value

Generalized Extreme Value

Papers

Fast parameter estimation of Generalized Extreme Value distribution using Neural Networks

Regression modelling of spatiotemporal extreme U.S. wildfires via partially-interpretable neural networks

DeepExtrema: A Deep Learning Approach for Forecasting Block Maxima in Time Series Data