Saved in:
Bibliographic Details
Main Authors: Dannemann, Joris, Junike, Gero
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2502.17913
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908242994528256
author Dannemann, Joris
Junike, Gero
author_facet Dannemann, Joris
Junike, Gero
contents Batch normalization is one of the most important regularization techniques for neural networks, significantly improving training by centering the layers of the neural network. There have been several attempts to provide a theoretical justification for batch ormalization. Santurkar and Tsipras (2018) [How does batch normalization help optimization? Advances in neural information rocessing systems, 31] claim that batch normalization improves initialization. We provide a counterexample showing that this claim s not true, i.e., batch normalization does not improve initialization.
format Preprint
id arxiv_https___arxiv_org_abs_2502_17913
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Batch normalization does not improve initialization
Dannemann, Joris
Junike, Gero
Machine Learning
Probability
Batch normalization is one of the most important regularization techniques for neural networks, significantly improving training by centering the layers of the neural network. There have been several attempts to provide a theoretical justification for batch ormalization. Santurkar and Tsipras (2018) [How does batch normalization help optimization? Advances in neural information rocessing systems, 31] claim that batch normalization improves initialization. We provide a counterexample showing that this claim s not true, i.e., batch normalization does not improve initialization.
title Batch normalization does not improve initialization
topic Machine Learning
Probability
url https://arxiv.org/abs/2502.17913