Robustness of Deep Neural Networks for Micro-Doppler Radar Classification

arXiv:2402.13651v1 Announce Type: cross
Abstract: 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 …

abstract architectures arxiv capabilities classification classifiers cs.cv cs.lg data data processing dataset eess.sp features micro networks neural networks practice processing radar risks robustness standard training type work