Feature Collapse
Machine Learning
2023-05-26 v1 Artificial Intelligence
Computation and Language
Abstract
We formalize and study a phenomenon called feature collapse that makes precise the intuitive idea that entities playing a similar role in a learning task receive similar representations. As feature collapse requires a notion of task, we leverage a simple but prototypical NLP task to study it. We start by showing experimentally that feature collapse goes hand in hand with generalization. We then prove that, in the large sample limit, distinct words that play identical roles in this NLP task receive identical local feature representations in a neural network. This analysis reveals the crucial role that normalization mechanisms, such as LayerNorm, play in feature collapse and in generalization.
Cite
@article{arxiv.2305.16162,
title = {Feature Collapse},
author = {Thomas Laurent and James H. von Brecht and Xavier Bresson},
journal= {arXiv preprint arXiv:2305.16162},
year = {2023}
}