An official website of the United States government
Several recent works have demonstrated highly effective model stealing (MS) attacks on Deep Neural Networks (DNNs) in black-box settings, even when the training data is unavailable. These attacks typically use some form of Out of Distribution (OOD) data to query the target model and use the predictions obtained to train a clone model. Such a clone model learns to approximate the decision boundary of the target model, achieving high accuracy on in-distribution examples. We propose Ensemble of Diverse Models (EDM) to defend against such MS attacks. EDM is made up of models that are trained to produce dissimilar predictions for OOD inputs. By using a different member of the ensemble to service different queries, our defense produces predictions that are highly discontinuous in the input space for the adversary's OOD queries. Such discontinuities cause the clone model trained on these predictions to have poor generalization on in-distribution examples. Our evaluations on several image classification tasks demonstrate that EDM defense can severely degrade the accuracy of clone models (up to 39.7%). Our defense has minimal impact on the target accuracy, negligible computational costs during inference, and is compatible with existing defenses for MS attacks.
more »
« less
MAZE: Data-Free Model Stealing Attack Using Zeroth-Order Gradient Estimation
High quality Machine Learning (ML) models are often considered valuable intellectual property by companies. Model Stealing (MS) attacks allow an adversary with black-box access to a ML model to replicate its functionality by training a clone model using the predictions of the target model for different inputs. However, best available existing MS attacks fail to produce a high-accuracy clone without access to the target dataset or a representative dataset necessary to query the target model. In this paper, we show that preventing access to the target dataset is not an adequate defense to protect a model. We propose MAZE -- a data-free model stealing attack using zeroth-order gradient estimation that produces high-accuracy clones. In contrast to prior works, MAZE uses only synthetic data created using a generative model to perform MS. Our evaluation with four image classification models shows that MAZE provides a normalized clone accuracy in the range of 0.90x to 0.99x, and outperforms even the recent attacks that rely on partial data (JBDA, clone accuracy 0.13x to 0.69x) and on surrogate data (KnockoffNets, clone accuracy 0.52x to 0.97x). We also study an extension of MAZE in the partial-data setting and develop MAZE-PD, which generates synthetic data closer to the target distribution. MAZE-PD further improves the clone accuracy 0.97x to 1.0x) and reduces the query budget required for the attack by 2x-24x.
more »
« less
- Award ID(s):
- 1646392
- PAR ID:
- 10322049
- Date Published:
- Journal Name:
- Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Recent model-extraction attacks on Machine Learning as a Service (MLaaS) systems have moved towards data-free approaches, showing the feasibility of stealing models trained with difficult-to-access data. However, these attacks are ineffective or limited due to the low accuracy of extracted models and the high number of queries to the models under attack. The high query cost makes such techniques infeasible for online MLaaS systems that charge per query.We create a novel approach to get higher accuracy and query efficiency than prior data-free model extraction techniques. Specifically, we introduce a novel generator training scheme that maximizes the disagreement loss between two clone models that attempt to copy the model under attack. This loss, combined with diversity loss and experience replay, enables the generator to produce better instances to train the clone models. Our evaluation on popular datasets CIFAR-10 and CIFAR-100 shows that our approach improves the final model accuracy by up to 3.42% and 18.48% respectively. The average number of queries required to achieve the accuracy of the prior state of the art is reduced by up to 64.95%. We hope this will promote future work on feasible data-free model extraction and defenses against such attacks.more » « less
-
Model stealing attacks on AI/ML devices undermine intellectual property rights, compromise the competitive advantage of the original model developers, and potentially expose sensitive data embedded in the model’s behavior to unauthorized parties. While previous research works have demonstrated successful side-channelbased model recovery in embedded microcontrollers and FPGA-based accelerators, the exploration of attacks on commercial ML accelerators remains largely unexplored. Moreover, prior side-channel attacks fail when they encounter previously unknown models. This paper demonstrates the first successful model extraction attack on the Google Edge Tensor Processing Unit (TPU), an off-the-shelf ML accelerator. Specifically, we show a hyperparameter stealing attack that can extract all layer configurations including the layer type, number of nodes, kernel/filter sizes, number of filters, strides, padding, and activation function. Most notably, our attack is the first comprehensive attack that can extract previously unseen models. This is achieved through an online template-building approach instead of a pre-trained ML-based approach used in prior works. Our results on a black-box Google Edge TPU evaluation show that, through obtained electromagnetic traces, our proposed framework can achieve 99.91% accuracy, making it the most accurate one to date. Our findings indicate that attackers can successfully extract various types of models on a black-box commercial TPU with utmost detail and call for countermeasures.more » « less
-
Recent advancements in Deep Neural Networks (DNNs) have enabled widespread deployment in multiple security-sensitive domains. The need for resource-intensive training and the use of valuable domain-specific training data have made these models the top intellectual property (IP) for model owners. One of the major threats to DNN privacy is model extraction attacks where adversaries attempt to steal sensitive information in DNN models. In this work, we propose an advanced model extraction framework DeepSteal that steals DNN weights remotely for the first time with the aid of a memory side-channel attack. Our proposed DeepSteal comprises two key stages. Firstly, we develop a new weight bit information extraction method, called HammerLeak, through adopting the rowhammer-based fault technique as the information leakage vector. HammerLeak leverages several novel system-level techniques tailored for DNN applications to enable fast and efficient weight stealing. Secondly, we propose a novel substitute model training algorithm with Mean Clustering weight penalty, which leverages the partial leaked bit information effectively and generates a substitute prototype of the target victim model. We evaluate the proposed model extraction framework on three popular image datasets (e.g., CIFAR-10/100/GTSRB) and four DNN architectures (e.g., ResNet-18/34/Wide-ResNetNGG-11). The extracted substitute model has successfully achieved more than 90% test accuracy on deep residual networks for the CIFAR-10 dataset. Moreover, our extracted substitute model could also generate effective adversarial input samples to fool the victim model. Notably, it achieves similar performance (i.e., ~1-2% test accuracy under attack) as white-box adversarial input attack (e.g., PGD/Trades).more » « less
-
In a black-box setting, the adversary only has API access to the target model and each query is expensive. Prior work on black-box adversarial examples follows one of two main strategies: (1) transfer attacks use white-box attacks on local models to find candidate adversarial examples that transfer to the target model, and (2) optimization-based attacks use queries to the target model and apply optimization techniques to search for adversarial examples. We propose hybrid attacks that combine both strategies, using candidate adversarial examples from local models as starting points for optimization-based attacks and using labels learned in optimization-based attacks to tune local models for finding transfer candidates. We empirically demonstrate on the MNIST, CIFAR10, and ImageNet datasets that our hybrid attack strategy reduces cost and improves success rates, and in combination with our seed prioritization strategy, enables batch attacks that can efficiently find adversarial examples with only a handful of queries.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- https://doi.org/10.1109/CVPR46437.2021.01360
