Security Analysis of Deep Neural Networks Operating in the Presence of Cache Side-Channel Attacks

Abstract

Recent work has introduced attacks that extract the architecture information of deep neural networks (DNN), as this knowledge enhances an adversary's capability to conduct black-box attacks against the model. This paper presents the first in-depth security analysis of DNN fingerprinting attacks that exploit cache side-channels. First, we define the threat model for these attacks: our adversary does not need the ability to query the victim model; instead, she runs a co-located process on the host machine victim's deep learning (DL) system is running and passively monitors the accesses of the target functions in the shared framework. Second, we introduce DeepRecon, an attack that reconstructs the architecture of the victim network by using the internal information extracted via Flush+Reload, a cache side-channel technique. Once the attacker observes function invocations that map directly to architecture attributes of the victim network, the attacker can reconstruct the victim's entire network architecture. In our evaluation, we demonstrate that an attacker can accurately reconstruct two complex networks (VGG19 and ResNet50) having observed only one forward propagation. Based on the extracted architecture attributes, we also demonstrate that an attacker can build a meta-model that accurately fingerprints the architecture and family of the pre-trained model in a transfer learning setting. From this meta-model, we evaluate the importance of the observed attributes in the fingerprinting process. Third, we propose and evaluate new framework-level defense techniques that obfuscate our attacker's observations. Our empirical security analysis represents a step toward understanding the DNNs' vulnerability to cache side-channel attacks.

References in corpus (9)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
• Ensemble Adversarial Training: Attacks and Defenses
• Resnet in Resnet: Generalizing Residual Architectures
• The Space of Transferable Adversarial Examples
• Black-box Adversarial Attacks with Limited Queries and Information
• Understanding Membership Inferences on Well-Generalized Learning Models
• Stealing Hyperparameters in Machine Learning
• Cache Telepathy: Leveraging Shared Resource Attacks to Learn DNN Architectures

Cited by in corpus (10)

• Stealing Neural Networks via Timing Side Channels
• High Accuracy and High Fidelity Extraction of Neural Networks
• Physical Side-Channel Attacks on Embedded Neural Networks: A Survey
• A framework for the extraction of Deep Neural Networks by leveraging public data
• How to 0wn NAS in Your Spare Time
• Careful What You Wish For: on the Extraction of Adversarially Trained Models
• Stealing Deep Reinforcement Learning Models for Fun and Profit
• SpecuSym: Speculative Symbolic Execution for Cache Timing Leak Detection
• Database Reconstruction from Noisy Volumes: A Cache Side-Channel Attack on SQLite
• LEASH: Enhancing Micro-architectural Attack Detection with a Reactive Process Scheduler