Reference
Neural Collapse Under MSE Loss: Proximity to and Dynamics on the Central Path, X. Y. Han, Vardan Papyan, David L. Donoho. (2022)

Abstract

The recently discovered Neural Collapse (NC) phenomenon occurs pervasively in today's deep net training paradigm of driving cross-entropy (CE) loss towards zero. During NC, last-layer features...

Content citing this item

Pill

Neural Collapse in deep classifiers during Terminal Phase of Training

Neural collapse refers to the observation that the last two layers of neural networks that were trained for a long time take a very simple, …

Michael Panchenko
Robustness in ML
All works referenced in our site...