State at the time of submission: here
Implementation of
- Heteroscedastic aleatoric uncertainty as described in the paper What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision?.
- Epistemic uncertainty using deep ensembles from Simple and Scalable Predictive Uncertainty Estimation using Deep Ensembles
- Combined aleatoric and epistemic uncertainty using the approach described in What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision?, but with ensembles for epistemic uncertainty (and obviously not "in one model" because this implementation uses deep ensembles)
marcobellagente93/Bayesian_Regression was used as reference for this notebook. The dataset generation is the same as in the notebook, and the plots are also inspired by it.