Fractal and Regular Geometry of Deep Neural Networks

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Di Lillo, Simmaco, Marinucci, Domenico, Salvi, Michele, Vigogna, Stefano
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911403057610752
author Di Lillo, Simmaco
Marinucci, Domenico
Salvi, Michele
Vigogna, Stefano
author_facet Di Lillo, Simmaco
Marinucci, Domenico
Salvi, Michele
Vigogna, Stefano
contents We study the geometric properties of random neural networks by investigating the boundary volumes of their excursion sets for different activation functions, as the depth increases. More specifically, we show that, for activations which are not very regular (e.g., the Heaviside step function), the boundary volumes exhibit fractal behavior, with their Hausdorff dimension monotonically increasing with the depth. On the other hand, for activations which are more regular (e.g., ReLU, logistic and $\tanh$), as the depth increases, the expected boundary volumes can either converge to zero, remain constant or diverge exponentially, depending on a single spectral parameter which can be easily computed. Our theoretical results are confirmed in some numerical experiments based on Monte Carlo simulations.
format Preprint
id arxiv_https___arxiv_org_abs_2504_06250
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Fractal and Regular Geometry of Deep Neural Networks
Di Lillo, Simmaco
Marinucci, Domenico
Salvi, Michele
Vigogna, Stefano
Probability
Machine Learning
60G60, 62B10, 62M45, 68T07
We study the geometric properties of random neural networks by investigating the boundary volumes of their excursion sets for different activation functions, as the depth increases. More specifically, we show that, for activations which are not very regular (e.g., the Heaviside step function), the boundary volumes exhibit fractal behavior, with their Hausdorff dimension monotonically increasing with the depth. On the other hand, for activations which are more regular (e.g., ReLU, logistic and $\tanh$), as the depth increases, the expected boundary volumes can either converge to zero, remain constant or diverge exponentially, depending on a single spectral parameter which can be easily computed. Our theoretical results are confirmed in some numerical experiments based on Monte Carlo simulations.
title Fractal and Regular Geometry of Deep Neural Networks
topic Probability
Machine Learning
60G60, 62B10, 62M45, 68T07
url https://arxiv.org/abs/2504.06250