Stealing Neural Networks via Timing Side Channels

Abstract

Deep learning is gaining importance in many applications. However, Neural Networks face several security and privacy threats. This is particularly significant in the scenario where Cloud infrastructures deploy a service with Neural Network model at the back end. Here, an adversary can extract the Neural Network parameters, infer the regularization hyperparameter, identify if a data point was part of the training data, and generate effective transferable adversarial examples to evade classifiers. This paper shows how a Neural Network model is susceptible to timing side channel attack. In this paper, a black box Neural Network extraction attack is proposed by exploiting the timing side channels to infer the depth of the network. Although, constructing an equivalent architecture is a complex search problem, it is shown how Reinforcement Learning with knowledge distillation can effectively reduce the search space to infer a target model. The proposed approach has been tested with VGG architectures on CIFAR10 data set. It is observed that it is possible to reconstruct substitute models with test accuracy close to the target models and the proposed approach is scalable and independent of type of Neural Network architectures.

References in corpus (9)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Distilling the Knowledge in a Neural Network
• Neural Architecture Search with Reinforcement Learning
• Do Deep Nets Really Need to be Deep?
• Copycat CNN: Stealing Knowledge by Persuading Confession with Random Non-Labeled Data
• Security Analysis of Deep Neural Networks Operating in the Presence of Cache Side-Channel Attacks
• Cache Telepathy: Leveraging Shared Resource Attacks to Learn DNN Architectures
• PRADA: Protecting against DNN Model Stealing Attacks
• Revisiting Definitional Foundations of Oblivious RAM for Secure Processor Implementations

Cited by in corpus (17)

• Adversarial Frontier Stitching for Remote Neural Network Watermarking
• A framework for the extraction of Deep Neural Networks by leveraging public data
• Hermes Attack: Steal DNN Models with Lossless Inference Accuracy
• NeurObfuscator: A Full-stack Obfuscation Tool to Mitigate Neural Architecture Stealing
• How to 0wn NAS in Your Spare Time
• Confidential Machine Learning Computation in Untrusted Environments: A Systems Security Perspective
• AUTOLYCUS: Exploiting Explainable AI (XAI) for Model Extraction Attacks against Interpretable Models
• BoMaNet: Boolean Masking of an Entire Neural Network
• Can one hear the shape of a neural network?: Snooping the GPU via Magnetic Side Channel
• Stealing Deep Reinforcement Learning Models for Fun and Profit
• Open DNN Box by Power Side-Channel Attack
• SpecuSym: Speculative Symbolic Execution for Cache Timing Leak Detection
• Monitoring-based Differential Privacy Mechanism Against Query-Flooding Parameter Duplication Attack
• Trustworthy AI Inference Systems: An Industry Research View
• Mitigating Query-Flooding Parameter Duplication Attack on Regression Models with High-Dimensional Gaussian Mechanism
• De-Anonymizing Text by Fingerprinting Language Generation
• Power-Based Attacks on Spatial DNN Accelerators