Bandwidth Utilization Side-Channel On ML Inference Accelerators

Abstract—Accelerators used for machine learning (ML) inference provide great performance benefits over CPUs. Securing confidential model in inference against off-chip side-channel attacks is critical in harnessing the performance advantage in practice. Data and memory address encryption has been recently proposed to defend against off-chip attacks. In this paper, we demonstrate that bandwidth... » read more

Hardware Security For AI Accelerators

Dedicated accelerator hardware for artificial intelligence and machine learning (AI/ML) algorithms are increasingly prevalent in data centers and endpoint devices. These accelerators handle valuable data and models, and face a growing threat landscape putting AI/ML assets at risk. Using fundamental cryptographic security techniques performed by a hardware root of trust can safeguard these as... » read more

AI Chip Architectures Race To The Edge

As machine-learning apps start showing up in endpoint devices and along the network edge of the IoT, the accelerators that make AI possible may look more like FPGA and SoC modules than current data-center-bound chips from Intel or Nvidia. Artificial intelligence and machine learning need powerful chips for computing answers (inference) from large data sets (training). Most AI chips—both tr... » read more