Impact of internal noise on convolutional neural networks

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Kolesnikov, Ivan, Semenova, Nadezhda
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866911018195615744
author Kolesnikov, Ivan
Semenova, Nadezhda
author_facet Kolesnikov, Ivan
Semenova, Nadezhda
contents In this paper, we investigate the impact of noise on a simplified trained convolutional network. The types of noise studied originate from a real optical implementation of a neural network, but we generalize these types to enhance the applicability of our findings on a broader scale. The noise types considered include additive and multiplicative noise, which relate to how noise affects individual neurons, as well as correlated and uncorrelated noise, which pertains to the influence of noise across one layers. We demonstrate that the propagation of uncorrelated noise primarily depends on the statistical properties of the connection matrices. Specifically, the mean value of the connection matrix following the layer impacted by noise governs the propagation of correlated additive noise, while the mean of its square contributes to the accumulation of uncorrelated noise. Additionally, we propose an analytical assessment of the noise level in the network's output signal, which shows a strong correlation with the results of numerical simulations.
format Preprint
id arxiv_https___arxiv_org_abs_2505_06611
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Impact of internal noise on convolutional neural networks
Kolesnikov, Ivan
Semenova, Nadezhda
Adaptation and Self-Organizing Systems
Neural and Evolutionary Computing
82C32, 68Txx, 60H40
In this paper, we investigate the impact of noise on a simplified trained convolutional network. The types of noise studied originate from a real optical implementation of a neural network, but we generalize these types to enhance the applicability of our findings on a broader scale. The noise types considered include additive and multiplicative noise, which relate to how noise affects individual neurons, as well as correlated and uncorrelated noise, which pertains to the influence of noise across one layers. We demonstrate that the propagation of uncorrelated noise primarily depends on the statistical properties of the connection matrices. Specifically, the mean value of the connection matrix following the layer impacted by noise governs the propagation of correlated additive noise, while the mean of its square contributes to the accumulation of uncorrelated noise. Additionally, we propose an analytical assessment of the noise level in the network's output signal, which shows a strong correlation with the results of numerical simulations.
title Impact of internal noise on convolutional neural networks
topic Adaptation and Self-Organizing Systems
Neural and Evolutionary Computing
82C32, 68Txx, 60H40
url https://arxiv.org/abs/2505.06611