Likelihood-Free Parameter Estimation with Neural Bayes Estimators
Sainsbury-Dale, Matthew, Zammit-Mangion, Andrew, Huser, Raphaël
Neural point estimators are neural networks that map data to parameter point estimates. They are fast, likelihood free and, due to their amortised nature, amenable to fast bootstrap-based uncertainty quantification. In this paper, we aim to increase the awareness of statisticians to this relatively new inferential tool, and to facilitate its adoption by providing user-friendly open-source software. We also give attention to the ubiquitous problem of making inference from replicated data, which we address in the neural setting using permutation-invariant neural networks. Through extensive simulation studies we show that these neural point estimators can quickly and optimally (in a Bayes sense) estimate parameters in weakly-identified and highly-parameterised models with relative ease. We demonstrate their applicability through an analysis of extreme sea-surface temperature in the Red Sea where, after training, we obtain parameter estimates and bootstrap-based confidence intervals from hundreds of spatial fields in a fraction of a second.
Oct-4-2023
- Country:
- Indian Ocean > Red Sea (0.25)
- Oceania > Australia
- New South Wales > Wollongong (0.04)
- North America > United States
- New York > New York County
- New York City (0.04)
- New Jersey > Hudson County
- Hoboken (0.04)
- Massachusetts > Middlesex County
- Cambridge (0.04)
- Florida > Palm Beach County
- Boca Raton (0.04)
- New York > New York County
- Europe
- Austria > Vienna (0.14)
- United Kingdom > England
- Cambridgeshire > Cambridge (0.14)
- Switzerland > Vaud
- Lausanne (0.04)
- Asia > Middle East
- Saudi Arabia (0.35)
- Yemen (0.25)
- Africa
- Sudan (0.35)
- Eritrea (0.35)
- Middle East
- Genre:
- Research Report > New Finding (0.67)
- Industry:
- Education (0.67)
- Technology: