ImpNet: Imperceptible and blackbox-undetectable backdoors in compiled neural networks

Abstract

Early backdoor attacks against machine learning set off an arms race in attack and defence development. Defences have since appeared demonstrating some ability to detect backdoors in models or even remove them. These defences work by inspecting the training data, the model, or the integrity of the training procedure. In this work, we show that backdoors can be added during compilation, circumventing any safeguards in the data preparation and model training stages. As an illustration, the attacker can insert weight-based backdoors during the hardware compilation step that will not be detected by any training or data-preparation process. Next, we demonstrate that some backdoors, such as ImpNet, can only be reliably detected at the stage where they are inserted and removing them anywhere else presents a significant challenge. We conclude that machine-learning model security requires assurance of provenance along the entire technical pipeline, including the data, model architecture, compiler, and hardware specification.

Infectious Hospital Area, the First Affiliated Hospital of University of Science and Technology of China, Hefei, China

“…It is worth noting that the poisoned structure is only used during the inference process and will not affect the training process. Backdoor attacks based on poisoned control flow [12,34,53] occur when the victim model is deployed and run as part of software. Attackers insert new node modules into the control flow to inject a backdoor.…”