WSAM: Visual Explanations from Style Augmentation as Adversarial Attacker and Their Influence in Image Classification

Felipe Moreno-Vera, Edgar Medina, Jorge Poco

2023

Abstract

Currently, style augmentation is capturing attention due to convolutional neural networks (CNN) being strongly biased toward recognizing textures rather than shapes. Most existing styling methods either perform a low-fidelity style transfer or a weak style representation in the embedding vector. This paper outlines a style augmentation algorithm using stochastic-based sampling with noise addition for randomization improvement on a general linear transformation for style transfer. With our augmentation strategy, all models not only present incredible robustness against image stylizing but also outperform all previous methods and surpass the state-of-the-art performance for the STL-10 dataset. In addition, we present an analysis of the model interpretations under different style variations. At the same time, we compare comprehensive experiments demonstrating the performance when applied to deep neural architectures in training settings.

Paper Citation

in Harvard Style

Moreno-Vera F., Medina E. and Poco J. (2023). WSAM: Visual Explanations from Style Augmentation as Adversarial Attacker and Their Influence in Image Classification. In Proceedings of the 18th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2023) - Volume 5: VISAPP; ISBN 978-989-758-634-7, SciTePress, pages 830-837. DOI: 10.5220/0011795400003417

in Bibtex Style

@conference{visapp23,
author={Felipe Moreno-Vera and Edgar Medina and Jorge Poco},
title={WSAM: Visual Explanations from Style Augmentation as Adversarial Attacker and Their Influence in Image Classification},
booktitle={Proceedings of the 18th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2023) - Volume 5: VISAPP},
year={2023},
pages={830-837},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0011795400003417},
isbn={978-989-758-634-7},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 18th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2023) - Volume 5: VISAPP
TI - WSAM: Visual Explanations from Style Augmentation as Adversarial Attacker and Their Influence in Image Classification
SN - 978-989-758-634-7
AU - Moreno-Vera F.
AU - Medina E.
AU - Poco J.
PY - 2023
SP - 830
EP - 837
DO - 10.5220/0011795400003417
PB - SciTePress