Heteroscedastic Multi-Output Deep Ensemble for 2D Transport¶

Notebook: HeteroscedasticMultiOutputDeepEnsemble_Transport2D_Tutorial.ipynb

This tutorial uses HeteroscedasticMultiOutputDeepEnsembleRegressor on a 2D advection-diffusion transport problem. The ensemble predicts both concentration and flux magnitude fields, together with per-pixel aleatoric noise.

Key ideas: - convolutional backbones inside a deep ensemble, - multi-output field prediction, - uncertainty increasing for high-frequency OOD plumes.

Primary references: - Nix, Weigend (1994), Estimating the Mean and Variance of the Target Probability Distribution. DOI: 10.1109/ICNN.1994.374138 - Kendall, Gal (2017), What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision? NeurIPS proceedings - Lakshminarayanan, Pritzel, Blundell (2017), Simple and Scalable Predictive Uncertainty Estimation using Deep Ensembles. NeurIPS proceedings