Search for: All records

Deep Neural Networks (DNN) are vulnerable to adversarial perturbations — small changes crafted deliberately on the input to mislead the model for wrong predictions. Adversarial attacks have disastrous consequences for deep learning empowered critical applications. Existing defense and detection techniques both require extensive knowledge of the model, testing inputs and even execution details. They are not viable for general deep learning implementations where the model internal is unknown, a common ‘black-box’ scenario for model users. Inspired by the fact that electromagnetic (EM) emanations of a model inference are dependent on both operations and data and may contain footprints of different input classes, we propose a framework, EMShepherd, to capture EM traces of model execution, perform processing on traces and exploit them for adversarial detection. Only benign samples and their EM traces are used to train the adversarial detector: a set of EM classifiers and class-specific unsupervised anomaly detectors. When the victim model system is under attack by an adversarial example, the model execution will be different from executions for the known classes, and the EM trace will be different. We demonstrate that our air-gapped EMShepherd can effectively detect different adversarial attacks on a commonly used FPGA deep learning accelerator for both Fashion MNIST and CIFAR-10 datasets. It achieves a detection rate on most types of adversarial samples, which is comparable to the state-of-the-art ‘white-box’ software-based detectors. 

Architecture reverse engineering has become an emerging attack against deep neural network (DNN) implemen- tations. Several prior works have utilized side-channel leakage to recover the model architecture while the an DNN is executing on a hardware acceleration platform. In this work, we target an open- source deep-learning accelerator, Versatile Tensor Accelerator (VTA), and utilize electromagnetic (EM) side-channel leakage to comprehensively learn the association between DNN architecture configurations and EM emanations. We also consider the holistic system – including the low-level tensor program code of the VTA accelerator on a Xilinx FPGA, and explore the effect of such low- level configurations on the EM leakage. Our study demonstrates that both the optimization and configuration of tensor programs will affect the EM side-channel leakage. Gaining knowledge of the association between low-level tensor program and the EM emanations, we propose NNReArch, a lightweight tensor program scheduling framework against side- channel-based DNN model architecture reverse engineering. Specifically, NNReArch targets reshaping the EM traces of different DNN operators, through scheduling the tensor program execution of the DNN model so as to confuse the adversary. NNReArch is a comprehensive protection framework supporting two modes, a balanced mode that strikes a balance between the DNN model confidentiality and execution performance, and a secure mode where the most secure setting is chosen. We imple- ment and evaluate the proposed framework on the open-source VTA with state-of-the-art DNN architectures. The experimental results demonstrate that NNReArch can efficiently enhance the model architecture security with a small performance overhead. In addition, the proposed obfuscation technique makes reverse engineering of the DNN architecture significantly harder. 

Abstract Recent advances in machine learning have enabled Neural Network (NN) inference directly on constrained embedded devices. This local approach enhances the privacy of user data, as the inputs to the NN inference are not shared with third-party cloud providers over a communication network. At the same time, however, performing local NN inference on embedded devices opens up the possibility of Power Analysis attacks, which have recently been shown to be effective in recovering NN parameters, as well as their activations and structure. Knowledge of these NN characteristics constitutes a privacy threat, as it enables highly effective Membership Inference and Model Inversion attacks, which can recover information about the sensitive data that the NN model was trained on. In this paper we address the problem of securing sensitive NN inference parameters against Power Analysis attacks. Our approach employs masking , a countermeasure well-studied in the context of cryptographic algorithms. We design a set of gadgets , i.e., masked operations, tailored to NN inference. We prove our proposed gadgets secure against power attacks and show, both formally and experimentally, that they are composable, resulting in secure NN inference. We further propose optimizations that exploit intrinsic characteristics of NN inference to reduce the masking’s runtime and randomness requirements. We empirically evaluate the performance of our constructions, showing them to incur a slowdown by a factor of about 2–5. 

This paper proposes Characteristic Examples for effectively fingerprinting deep neural networks, featuring high-robustness to the base model against model pruning as well as low-transferability to unassociated models. This is the first work taking both robustness and transferability into consideration for generating realistic fingerprints, whereas current methods lack practical assumptions and may incur large false positive rates. To achieve better trade-off between robustness and transferability, we propose three kinds of characteristic examples: vanilla C-examples, RC-examples, and LTRC-example, to derive fingerprints from the original base model. To fairly characterize the trade-off between robustness and transferability, we propose Uniqueness Score, a comprehensive metric that measures the difference between robustness and transferability, which also serves as an indicator to the false alarm problem. Extensive experiments demonstrate that the proposed characteristic examples can achieve superior performance when compared with existing fingerprinting methods. In particular, for VGG ImageNet models, using LTRC-examples gives 4X higher uniqueness score than the baseline method and does not incur any false positives.