The Role of Noisy Data in Improving CNN Robustness for Image Classification

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Ramírez-Agudelo, Oscar H., Gorea, Nicoleta, Reif, Aliza, Bonasera, Lorenzo, Karl, Michael
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866908761321373696
author Ramírez-Agudelo, Oscar H.
Gorea, Nicoleta
Reif, Aliza
Bonasera, Lorenzo
Karl, Michael
author_facet Ramírez-Agudelo, Oscar H.
Gorea, Nicoleta
Reif, Aliza
Bonasera, Lorenzo
Karl, Michael
contents Data quality plays a central role in the performance and robustness of convolutional neural networks (CNNs) for image classification. While high-quality data is often preferred for training, real-world inputs are frequently affected by noise and other distortions. This paper investigates the effect of deliberately introducing controlled noise into the training data to improve model robustness. Using the CIFAR-10 dataset, we evaluate the impact of three common corruptions, namely Gaussian noise, Salt-and-Pepper noise, and Gaussian blur at varying intensities and training set pollution levels. Experiments using a Resnet-18 model reveal that incorporating just 10\% noisy data during training is sufficient to significantly reduce test loss and enhance accuracy under fully corrupted test conditions, with minimal impact on clean-data performance. These findings suggest that strategic exposure to noise can act as a simple yet effective regularizer, offering a practical trade-off between traditional data cleanliness and real-world resilience.
format Preprint
id arxiv_https___arxiv_org_abs_2601_08043
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle The Role of Noisy Data in Improving CNN Robustness for Image Classification
Ramírez-Agudelo, Oscar H.
Gorea, Nicoleta
Reif, Aliza
Bonasera, Lorenzo
Karl, Michael
Computer Vision and Pattern Recognition
Artificial Intelligence
Data quality plays a central role in the performance and robustness of convolutional neural networks (CNNs) for image classification. While high-quality data is often preferred for training, real-world inputs are frequently affected by noise and other distortions. This paper investigates the effect of deliberately introducing controlled noise into the training data to improve model robustness. Using the CIFAR-10 dataset, we evaluate the impact of three common corruptions, namely Gaussian noise, Salt-and-Pepper noise, and Gaussian blur at varying intensities and training set pollution levels. Experiments using a Resnet-18 model reveal that incorporating just 10\% noisy data during training is sufficient to significantly reduce test loss and enhance accuracy under fully corrupted test conditions, with minimal impact on clean-data performance. These findings suggest that strategic exposure to noise can act as a simple yet effective regularizer, offering a practical trade-off between traditional data cleanliness and real-world resilience.
title The Role of Noisy Data in Improving CNN Robustness for Image Classification
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2601.08043