Announcement_24

Created in August 10, 2026

2026

New preprint Tracing sources of epistemic uncertainty in deep learning predictions: homo- and hetero-scedastic linearized estimators of our work on sources of uncertainty in deep learning with Pierre Nodet.

---

Enjoy Reading This Article?

Here are some more articles you might like to read next:

Derivatives through a batch norm layer

What is the empirical Fisher ?

How to compute the Fisher of a conditional when applying natural gradient to neural networks?

The algebra of second order methods in neural networks

Demystifying Natural Neural Networks