Sinh-arcsinh-normal distributions to add uncertainty to neural network regression tasks: Applications to tropical cyclone intensity forecasts
Open Access
- 15 June 2023
- journal article
- research article
- Published by Cambridge University Press (CUP) in Environmental Data Science
- Vol. 2, e15
Abstract
A simple method for adding uncertainty to neural network regression tasks in earth science via estimation of a general probability distribution is described. Specifically, we highlight the sinh-arcsinh-normal distributions as particularly well suited for neural network uncertainty estimation. The methodology supports estimation of heteroscedastic, asymmetric uncertainties by a simple modification of the network output and loss function. Method performance is demonstrated by predicting tropical cyclone intensity forecast uncertainty and by comparing two other common methods for neural network uncertainty quantification (i.e., Bayesian neural networks and Monte Carlo dropout). The simple approach described here is intuitive and applicable when no prior exists and one just wishes to parameterize the output and its uncertainty according to some previously defined family of distributions. The authors believe it will become a powerful, go-to method moving forward.Keywords
Funding Information
- Directorate for Geosciences (AGS-1749261)
This publication has 28 references indexed in Scilit:
- Reliable probabilistic forecasts from an ensemble reservoir inflow forecasting systemWater Resources Research, 2014
- Probabilistic ForecastingAnnual Review of Statistics and Its Application, 2014
- Atlantic Hurricane Database Uncertainty and Presentation of a New Database FormatMonthly Weather Review, 2013
- Neural Networks2012
- Calibrating probabilistic forecasts from an NWP ensembleTellus, 2011
- Sinh-arcsinh distributionsBiometrika, 2009
- Probabilistic Forecasts, Calibration and SharpnessJournal of the Royal Statistical Society Series B: Statistical Methodology, 2007
- Interpretation of Rank Histograms for Verifying Ensemble ForecastsMonthly Weather Review, 2001
- Present Position and Potential Developments: Some Personal Views: Statistical Theory: The Prequential ApproachJournal of the Royal Statistical Society. Series A (General), 1984
- A generalized probability density function for double-bounded random processesJournal of Hydrology, 1980