Please use this identifier to cite or link to this item:
|Title:||DeepFreeze : cold boot attacks and high fidelity model recovery on commercial EdgeML device||Authors:||Won, Yoo-Seung
|Keywords:||Engineering::Electrical and electronic engineering::Computer hardware, software and systems||Issue Date:||2021||Source:||Won, Y., Chatterjee, S., Jap, D., Basu, A. & Bhasin, S. (2021). DeepFreeze : cold boot attacks and high fidelity model recovery on commercial EdgeML device. 2021 IEEE/ACM International Conference On Computer Aided Design (ICCAD), 1-9. https://dx.doi.org/10.1109/ICCAD51958.2021.9643512||Project:||NRF2018NCR- NCR009-0001||Abstract:||EdgeML accelerators like Intel Neural Compute Stick 2 (NCS) can enable efficient edge-based inference with complex pre-trained models. The models are loaded in the host (like Raspberry Pi) and then transferred to NCS for inference. In this paper, we demonstrate practical and low-cost cold boot based model recovery attacks on NCS to recover the model architecture and weights, loaded from the Raspberry Pi. The architecture is recovered with 100% success and weights with an error rate of 0.04%. The recovered model reports maximum accuracy loss of 0.5% as compared to original model and allows high fidelity transfer of adversarial examples. We further extend our study to other cold boot attack setups reported in the literature with higher error rates leading to accuracy loss as high as 70%. We then propose a methodology based on knowledge distillation to correct the erroneous weights in recovered model, even without access to original training data. The proposed attack remains unaffected by the model encryption features of the OpenVINO and NCS framework.||URI:||https://hdl.handle.net/10356/156094||ISBN:||9781665445078||DOI:||10.1109/ICCAD51958.2021.9643512||Rights:||© 2021 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. The published version is available at: https://doi.org/10.1109/ICCAD51958.2021.9643512.||Fulltext Permission:||open||Fulltext Availability:||With Fulltext|
|Appears in Collections:||EEE Conference Papers|
TL Conference Papers
Updated on May 17, 2022
Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.