A hyperspectral adversarial attack that combines block-wise spatial-spectral transformations with a variance-weighted feature divergence loss improves transferability over MI-FGSM baselines.
Convolutional neural networks for hyperspec- tral image classification,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
support 1representative citing papers
citing papers explorer
-
Boosting Adversarial Transferability for Hyperspectral Image Classification Using 3D Structure-invariant Transformation and Weighted Intermediate Feature Divergence
A hyperspectral adversarial attack that combines block-wise spatial-spectral transformations with a variance-weighted feature divergence loss improves transferability over MI-FGSM baselines.