A new adversarial attack method for medical hyperspectral imaging exploits spectral-spatial dependencies and multiscale features to degrade classification performance more effectively in tumor regions than prior baselines.
Attacking cnns in histopathology with snap: sporadic and naturalistic adversarial patches (student abstract),
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Adversarial Attacks on Medical Hyperspectral Imaging Exploiting Spectral-Spatial Dependencies and Multiscale Features
A new adversarial attack method for medical hyperspectral imaging exploits spectral-spatial dependencies and multiscale features to degrade classification performance more effectively in tumor regions than prior baselines.