all InfoSec news
Evil from Within: Machine Learning Backdoors through Hardware Trojans. (arXiv:2304.08411v2 [cs.CR] UPDATED)
cs.CR updates on arXiv.org arxiv.org
Backdoors pose a serious threat to machine learning, as they can compromise
the integrity of security-critical systems, such as self-driving cars. While
different defenses have been proposed to address this threat, they all rely on
the assumption that the hardware on which the learning models are executed
during inference is trusted. In this paper, we challenge this assumption and
introduce a backdoor attack that completely resides within a common hardware
accelerator for machine learning. Outside of the accelerator, neither the …
accelerator address attack backdoor backdoors cars challenge compromise critical critical systems driving hardware integrity machine machine learning security self-driving self-driving cars serious software systems threat trojans