Lie to Me: A Soft Threshold Defense Method for Adversarial Examples of Remote Sensing Images

• Published in: IEEE geoscience and remote sensing letters Vol. 19; pp. 1 - 5
• Main Authors: Chen, Li, Xiao, Jun, Zou, Pu, Li, Haifeng
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 2022; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Adversarial example; Algorithms; Artificial neural networks; Computational modeling; convolutional neural network (CNN); deep learning; Image classification; Logistics; Mathematical model; Neural networks; Perturbation; Perturbation methods; Prediction algorithms; Predictive models; Remote sensing; Selectivity; Vulnerability

Adversarial examples fool the models into predicting wrong results through generated perturbations, demonstrating the vulnerability of convolutional neural networks (CNNs). Recent studies also show that many CNNs applied to remote sensing image (RSI) scene classification are still subject to adversa...