Analyzing the Adversarial Robustness and Interpretability of Deep SAR Classification Models: A Comprehensive Examination of Their Reliability

Deep neural networks (DNNs) have shown strong performance in synthetic aperture radar (SAR) image classification. However, their “black-box” nature limits interpretability and poses challenges for robustness, which is critical for sensitive applications such as disaster assessment, environmental mon...

Full description

Saved in:
Bibliographic Details
Main Authors: Tianrui Chen, Limeng Zhang, Weiwei Guo, Zenghui Zhang, Mihai Datcu
Format: Article
Language:English
Published: MDPI AG 2025-06-01
Series:Remote Sensing
Subjects:
Online Access:https://www.mdpi.com/2072-4292/17/11/1943
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849721863485259776
author Tianrui Chen
Limeng Zhang
Weiwei Guo
Zenghui Zhang
Mihai Datcu
author_facet Tianrui Chen
Limeng Zhang
Weiwei Guo
Zenghui Zhang
Mihai Datcu
author_sort Tianrui Chen
collection DOAJ
description Deep neural networks (DNNs) have shown strong performance in synthetic aperture radar (SAR) image classification. However, their “black-box” nature limits interpretability and poses challenges for robustness, which is critical for sensitive applications such as disaster assessment, environmental monitoring, and agricultural insurance. This study systematically evaluates the adversarial robustness of five representative DNNs (VGG11/16, ResNet18/101, and A-ConvNet) under a variety of attack and defense settings. Using eXplainable AI (XAI) techniques and attribution-based visualizations, we analyze how adversarial perturbations and adversarial training affect model behavior and decision logic. Our results reveal significant robustness differences across architectures, highlight interpretability limitations, and suggest practical guidelines for building more robust SAR classification systems. We also discuss challenges associated with large-scale, multi-class land use and land cover (LULC) classification under adversarial conditions.
format Article
id doaj-art-16ee39dee6054916aa9b734af2bbbd7f
institution DOAJ
issn 2072-4292
language English
publishDate 2025-06-01
publisher MDPI AG
record_format Article
series Remote Sensing
spelling doaj-art-16ee39dee6054916aa9b734af2bbbd7f2025-08-20T03:11:32ZengMDPI AGRemote Sensing2072-42922025-06-011711194310.3390/rs17111943Analyzing the Adversarial Robustness and Interpretability of Deep SAR Classification Models: A Comprehensive Examination of Their ReliabilityTianrui Chen0Limeng Zhang1Weiwei Guo2Zenghui Zhang3Mihai Datcu4Shanghai Key Laboratory of Intelligent Sensing and Recognition, Shanghai Jiao Tong University, Shanghai 200240, ChinaShanghai Key Laboratory of Intelligent Sensing and Recognition, Shanghai Jiao Tong University, Shanghai 200240, ChinaCenter of Digital Innovation, Tongji University, Shanghai 200092, ChinaShanghai Key Laboratory of Intelligent Sensing and Recognition, Shanghai Jiao Tong University, Shanghai 200240, ChinaResearch Center for Spatial Information (CEOSpaceTech), POLITEHNICA Bucharest, Bucharest 011061, RomaniaDeep neural networks (DNNs) have shown strong performance in synthetic aperture radar (SAR) image classification. However, their “black-box” nature limits interpretability and poses challenges for robustness, which is critical for sensitive applications such as disaster assessment, environmental monitoring, and agricultural insurance. This study systematically evaluates the adversarial robustness of five representative DNNs (VGG11/16, ResNet18/101, and A-ConvNet) under a variety of attack and defense settings. Using eXplainable AI (XAI) techniques and attribution-based visualizations, we analyze how adversarial perturbations and adversarial training affect model behavior and decision logic. Our results reveal significant robustness differences across architectures, highlight interpretability limitations, and suggest practical guidelines for building more robust SAR classification systems. We also discuss challenges associated with large-scale, multi-class land use and land cover (LULC) classification under adversarial conditions.https://www.mdpi.com/2072-4292/17/11/1943synthetic aperture radar (SAR)image classificationdeep learningadversarial exampleexplainable artificial intelligence
spellingShingle Tianrui Chen
Limeng Zhang
Weiwei Guo
Zenghui Zhang
Mihai Datcu
Analyzing the Adversarial Robustness and Interpretability of Deep SAR Classification Models: A Comprehensive Examination of Their Reliability
Remote Sensing
synthetic aperture radar (SAR)
image classification
deep learning
adversarial example
explainable artificial intelligence
title Analyzing the Adversarial Robustness and Interpretability of Deep SAR Classification Models: A Comprehensive Examination of Their Reliability
title_full Analyzing the Adversarial Robustness and Interpretability of Deep SAR Classification Models: A Comprehensive Examination of Their Reliability
title_fullStr Analyzing the Adversarial Robustness and Interpretability of Deep SAR Classification Models: A Comprehensive Examination of Their Reliability
title_full_unstemmed Analyzing the Adversarial Robustness and Interpretability of Deep SAR Classification Models: A Comprehensive Examination of Their Reliability
title_short Analyzing the Adversarial Robustness and Interpretability of Deep SAR Classification Models: A Comprehensive Examination of Their Reliability
title_sort analyzing the adversarial robustness and interpretability of deep sar classification models a comprehensive examination of their reliability
topic synthetic aperture radar (SAR)
image classification
deep learning
adversarial example
explainable artificial intelligence
url https://www.mdpi.com/2072-4292/17/11/1943
work_keys_str_mv AT tianruichen analyzingtheadversarialrobustnessandinterpretabilityofdeepsarclassificationmodelsacomprehensiveexaminationoftheirreliability
AT limengzhang analyzingtheadversarialrobustnessandinterpretabilityofdeepsarclassificationmodelsacomprehensiveexaminationoftheirreliability
AT weiweiguo analyzingtheadversarialrobustnessandinterpretabilityofdeepsarclassificationmodelsacomprehensiveexaminationoftheirreliability
AT zenghuizhang analyzingtheadversarialrobustnessandinterpretabilityofdeepsarclassificationmodelsacomprehensiveexaminationoftheirreliability
AT mihaidatcu analyzingtheadversarialrobustnessandinterpretabilityofdeepsarclassificationmodelsacomprehensiveexaminationoftheirreliability