Robust Fourier Neural Networks

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Jeong, Halyun, Han, Jihun
Format: Preprint
Publié: 2024
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866909303437262848
author Jeong, Halyun
Han, Jihun
author_facet Jeong, Halyun
Han, Jihun
contents Fourier embedding has shown great promise in removing spectral bias during neural network training. However, it can still suffer from high generalization errors, especially when the labels or measurements are noisy. We demonstrate that introducing a simple diagonal layer after the Fourier embedding layer makes the network more robust to measurement noise, effectively prompting it to learn sparse Fourier features. We provide theoretical justifications for this Fourier feature learning, leveraging recent developments in diagonal networks and implicit regularization in neural networks. Under certain conditions, our proposed approach can also learn functions that are noisy mixtures of nonlinear functions of Fourier features. Numerical experiments validate the effectiveness of our proposed architecture, supporting our theory.
format Preprint
id arxiv_https___arxiv_org_abs_2409_02052
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Robust Fourier Neural Networks
Jeong, Halyun
Han, Jihun
Machine Learning
65T40, 62J02, 68T07
Fourier embedding has shown great promise in removing spectral bias during neural network training. However, it can still suffer from high generalization errors, especially when the labels or measurements are noisy. We demonstrate that introducing a simple diagonal layer after the Fourier embedding layer makes the network more robust to measurement noise, effectively prompting it to learn sparse Fourier features. We provide theoretical justifications for this Fourier feature learning, leveraging recent developments in diagonal networks and implicit regularization in neural networks. Under certain conditions, our proposed approach can also learn functions that are noisy mixtures of nonlinear functions of Fourier features. Numerical experiments validate the effectiveness of our proposed architecture, supporting our theory.
title Robust Fourier Neural Networks
topic Machine Learning
65T40, 62J02, 68T07
url https://arxiv.org/abs/2409.02052