Doubting the weights: ensembles and calibration
A distribution instead of a single set of weights. Its practical counterpart is an ensemble, and when we measure it, it gives uncertainty in the right direction but at the wrong scale.
4 steps
175 XP
A free account is needed
Start the lesson →
Sources
- Lakshminarayanan, B., Pritzel, A. & Blundell, C. 2017 · Simple and Scalable Predictive Uncertainty Estimation Using Deep Ensembles · NeurIPS 2017
- Blundell, C. et al. 2015 · Weight Uncertainty in Neural Networks · ICML 2015
- Ovadia, Y. et al. 2019 · Can You Trust Your Model’s Uncertainty? · NeurIPS 2019
ML Academy · an interactive machine learning course that runs in your browser ·
All lessons