Robustness of Deep Neural Networks for Micro-Doppler Radar Classification

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Czerkawski, Mikolaj, Clemente, Carmine, Michie, Craig, Tachtatzis, Christos
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917595597242368
author Czerkawski, Mikolaj
Clemente, Carmine
Michie, Craig
Tachtatzis, Christos
author_facet Czerkawski, Mikolaj
Clemente, Carmine
Michie, Craig
Tachtatzis, Christos
contents With the great capabilities of deep classifiers for radar data processing come the risks of learning dataset-specific features that do not generalize well. In this work, the robustness of two deep convolutional architectures, trained and tested on the same data, is evaluated. When standard training practice is followed, both classifiers exhibit sensitivity to subtle temporal shifts of the input representation, an augmentation that carries minimal semantic content. Furthermore, the models are extremely susceptible to adversarial examples. Both small temporal shifts and adversarial examples are a result of a model overfitting on features that do not generalize well. As a remedy, it is shown that training on adversarial examples and temporally augmented samples can reduce this effect and lead to models that generalise better. Finally, models operating on cadence-velocity diagram representation rather than Doppler-time are demonstrated to be naturally more immune to adversarial examples.
format Preprint
id arxiv_https___arxiv_org_abs_2402_13651
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Robustness of Deep Neural Networks for Micro-Doppler Radar Classification
Czerkawski, Mikolaj
Clemente, Carmine
Michie, Craig
Tachtatzis, Christos
Computer Vision and Pattern Recognition
Machine Learning
Signal Processing
With the great capabilities of deep classifiers for radar data processing come the risks of learning dataset-specific features that do not generalize well. In this work, the robustness of two deep convolutional architectures, trained and tested on the same data, is evaluated. When standard training practice is followed, both classifiers exhibit sensitivity to subtle temporal shifts of the input representation, an augmentation that carries minimal semantic content. Furthermore, the models are extremely susceptible to adversarial examples. Both small temporal shifts and adversarial examples are a result of a model overfitting on features that do not generalize well. As a remedy, it is shown that training on adversarial examples and temporally augmented samples can reduce this effect and lead to models that generalise better. Finally, models operating on cadence-velocity diagram representation rather than Doppler-time are demonstrated to be naturally more immune to adversarial examples.
title Robustness of Deep Neural Networks for Micro-Doppler Radar Classification
topic Computer Vision and Pattern Recognition
Machine Learning
Signal Processing
url https://arxiv.org/abs/2402.13651