Reviewed Version (pdf): https://openreview.net/references/pdf?id=Ke1FBu2rL
Keywords: side channel, model extraction, GPU, magnetic induction, sensors
Abstract: We examine the magnetic flux emanating from a graphics processing unit’s (GPU’s) power cable, as acquired by a cheap $3 induction sensor, and find that this signal betrays the detailed topology and hyperparameters of a black-box neural network model. The attack acquires the magnetic signal for one query with unknown input values, but known input dimension and batch size. The reconstruction is possible due to the modular layer sequence in which deep neural networks are evaluated. We find that each layer component’s evaluation produces an identifiable magnetic signal signature, from which layer topology, width, function type, and sequence order can be inferred using a suitably trained classifier and an optimization based on integer programming. We study the extent to which network specifications can be recovered, and consider metrics for comparing network similarity. We demonstrate the potential accuracy of this side channel attack in recovering the details for a broad range of network architectures including also random designs. We consider applications that may exploit this novel side channel exposure, such as adversarial transfer attacks. In response, we discuss countermeasures to protect against our method and other similar snooping techniques.
One-sentence Summary: We examine the magnetic flux emanating from a graphics processing unit’s (GPU’s) power cable and find that this signal betrays the detailed topology and hyperparameters of a black-box neural network model.
Supplementary Material: zip
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics