Guardado en:
Detalles Bibliográficos
Autores principales: Pinto, Andrea, Rangamani, Akshay, Poggio, Tomaso
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:https://arxiv.org/abs/2411.13733
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866910707127156736
author Pinto, Andrea
Rangamani, Akshay
Poggio, Tomaso
author_facet Pinto, Andrea
Rangamani, Akshay
Poggio, Tomaso
contents While previous optimization results have suggested that deep neural networks tend to favour low-rank weight matrices, the implications of this inductive bias on generalization bounds remain underexplored. In this paper, we apply Maurer's chain rule for Gaussian complexity to analyze how low-rank layers in deep networks can prevent the accumulation of rank and dimensionality factors that typically multiply across layers. This approach yields generalization bounds for rank and spectral norm constrained networks. We compare our results to prior generalization bounds for deep networks, highlighting how deep networks with low-rank layers can achieve better generalization than those with full-rank layers. Additionally, we discuss how this framework provides new perspectives on the generalization capabilities of deep networks exhibiting neural collapse.
format Preprint
id arxiv_https___arxiv_org_abs_2411_13733
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle On Generalization Bounds for Neural Networks with Low Rank Layers
Pinto, Andrea
Rangamani, Akshay
Poggio, Tomaso
Machine Learning
While previous optimization results have suggested that deep neural networks tend to favour low-rank weight matrices, the implications of this inductive bias on generalization bounds remain underexplored. In this paper, we apply Maurer's chain rule for Gaussian complexity to analyze how low-rank layers in deep networks can prevent the accumulation of rank and dimensionality factors that typically multiply across layers. This approach yields generalization bounds for rank and spectral norm constrained networks. We compare our results to prior generalization bounds for deep networks, highlighting how deep networks with low-rank layers can achieve better generalization than those with full-rank layers. Additionally, we discuss how this framework provides new perspectives on the generalization capabilities of deep networks exhibiting neural collapse.
title On Generalization Bounds for Neural Networks with Low Rank Layers
topic Machine Learning
url https://arxiv.org/abs/2411.13733