Ravishka Shemal Rathnasuriya

AESOP introduces an adversarial attack that targets the entire execution path of deep learning pipelines, demonstrating that path-aware selection can inflate computational costs by orders of magnitude more than single-model attacks.