Shift-invariant universal adversarial attacks to avoid deep-learning-based modulation classification.
Keyu LuZhisheng QianManxi WangDewang WangPengfei MaPublished in: Int. J. Commun. Syst. (2023)
Keyphrases
- deep learning
- shift invariant
- unsupervised learning
- machine learning
- pattern recognition
- wavelet domain
- decision trees
- feature space
- supervised learning
- image classification
- sparse coding
- wavelet coefficients
- rotation invariant
- discrete wavelet transform
- text classification
- feature selection
- feature extraction
- level set
- feature set
- training set
- learning algorithm
- data mining