Mind control attack: Undermining deep learning with GPU memory exploitation

doi:10.1016/j.cose.2020.102115

Mind control attack: Undermining deep learning with GPU memory exploitation

2021-03-01

7

OA

AI

DOI：10.1016/j.cose.2020.102115

Original

Save

View PDF

Abstract

En 中文

Modern deep learning frameworks rely heavily on GPUs to accelerate the computation. However, the security implication of GPU device memory exploitation on deep learning frameworks has been largely neglected. In this paper, we argue that GPU device memory manipulation is a novel attack vector against deep learning systems. We present a novel attack method leveraging the attack vector, which makes deep learning predictions no longer different from random guessing by degrading the accuracy of the predictions. To the best of our knowledge, we are the first to show a practical attack that directly exploits deep learning frameworks through GPU memory manipulation. We confirmed that our attack works on three popular deep learning frameworks, TensorFlow, CNTK, and Caffe, running on CUDA. Finally, we propose potential defense mechanisms against our attack, and discuss concerns of GPU memory safety. (c) 2020 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/)

Keywords:

Graphics process unit security

GPU memory exploit

Deep learning security

Reverse engineering

Compute unified device architecture

Mind control attack: Undermining deep learning with GPU memory exploitation

Abstract

AI Summary

Journal

Organization