English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

With the increased usage of AI accelerators on mobile and edge devices,on-device machine learning (ML) is gaining popularity. Thousands of proprietaryML models are being deployed today on billions of untrusted devices. Thisraises serious security concerns about model privacy. However, protecting modelprivacy without losing access to the untrusted AI accelerators is a challengingproblem. In this paper, we present a novel on-device model inference system,ShadowNet. ShadowNet protects the model privacy with Trusted ExecutionEnvironment (TEE) while securely outsourcing the heavy linear layers of themodel to the untrusted hardware accelerators. ShadowNet achieves this bytransforming the weights of the linear layers before outsourcing them andrestoring the results inside the TEE. The non-linear layers are also keptsecure inside the TEE. ShadowNet's design ensures efficient transformation ofthe weights and the subsequent restoration of the results. We build a ShadowNetprototype based on TensorFlow Lite and evaluate it on five popular CNNs,namely, MobileNet, ResNet-44, MiniVGG, ResNet-404, and YOLOv4-tiny. Ourevaluation shows that ShadowNet achieves strong security guarantees withreasonable performance, offering a practical solution for secure on-devicemodel inference.

ShadowNet: A Secure and Efficient On-device Model Inference System for Convolutional Neural Networks