Neural networks : the official journal of the International Neural Network SocietyHumansDeep LearningImage ProcessingComputer-AssistedNeural Networks
Elevating adversarial robustness by contrastive multitasking defence in medical image segmentation.
Sneha Shukla, Puneet Gupta
Published: 202610.1016/j.neunet.2025.108182
Abstract
Although Deep Learning (DL)-based Medical Image Segmentation (MIS) models are critically important, adversarial attacks substantially diminish their efficacy. Such attacks subtly perturb inputs, causing the model to produce inaccurate predictions. Th…
Preview only. Read the full abstract at the source