Several years of research have shown that machine-learning systems are vulnerable to adversarial examples, both in theory and in practice. Until now, such attacks have primarily targeted visual models, exploiting the gap between human and machine perception. Although text-based models have also been attacked with adversarial examples, such attacks struggled to preserve semantic meaning and indistinguishability. In this paper, we explore a large class of adversarial examples that can be used to attack text-based models in a black-box setting without making any human-perceptible visual modification to inputs. We use encoding-specific perturbations that are imperceptible to the human eye to manipulate the outputs of a wide range of Natural Language Processing (NLP) systems from neural machine-translation pipelines to web search engines. We find that with a single imperceptible encoding injection -- representing one invisible character, homoglyph, reordering, or deletion -- an attacker can significantly reduce the performance of vulnerable models, and with three injections most models can be functionally broken. Our attacks work against currently-deployed commercial systems, including those produced by Microsoft and Google, in addition to open source models published by Facebook and IBM. This novel series of attacks presents a significant threat to many language processing systems: an attacker can affect systems in a targeted manner without any assumptions about the underlying model. We conclude that text-based NLP systems require careful input sanitization, just like conventional applications, and that given such systems are now being deployed rapidly at scale, the urgent attention of architects and operators is required.
Inpainting is a learned interpolation technique that is based on generative modeling and used to populate masked or missing pieces in an image; it has wide applications in picture editing and retouching. Recently, inpainting started being used for watermark removal, raising concerns. In this paper we study how to manipulate it using our markpainting technique. First, we show how an image owner with access to an inpainting model can augment their image in such a way that any attempt to edit it using that model will add arbitrary visible information. We find that we can target multiple different models simultaneously with our technique. This can be designed to reconstitute a watermark if the editor had been trying to remove it. Second, we show that our markpainting technique is transferable to models that have different architectures or were trained on different datasets, so watermarks created using it are difficult for adversaries to remove. Markpainting is novel and can be used as a manipulation alarm that becomes visible in the event of inpainting.
With increasingly more data and computation involved in their training, machine learning models constitute valuable intellectual property. This has spurred interest in model stealing, which is made more practical by advances in learning with partial, little, or no supervision. Existing defenses focus on inserting unique watermarks in a model's decision surface, but this is insufficient: the watermarks are not sampled from the training distribution and thus are not always preserved during model stealing. In this paper, we make the key observation that knowledge contained in the stolen model's training set is what is common to all stolen copies. The adversary's goal, irrespective of the attack employed, is always to extract this knowledge or its by-products. This gives the original model's owner a strong advantage over the adversary: model owners have access to the original training data. We thus introduce $dataset$ $inference$, the process of identifying whether a suspected model copy has private knowledge from the original model's dataset, as a defense against model stealing. We develop an approach for dataset inference that combines statistical testing with the ability to estimate the distance of multiple data points to the decision boundary. Our experiments on CIFAR10, SVHN, CIFAR100 and ImageNet show that model owners can claim with confidence greater than 99% that their model (or dataset as a matter of fact) was stolen, despite only exposing 50 of the stolen model's training points. Dataset inference defends against state-of-the-art attacks even when the adversary is adaptive. Unlike prior work, it does not require retraining or overfitting the defended model.
Machine learning is vulnerable to a wide variety of different attacks. It is now well understood that by changing the underlying data distribution, an adversary can poison the model trained with it or introduce backdoors. In this paper we present a novel class of training-time attacks that require no changes to the underlying model dataset or architecture, but instead only change the order in which data are supplied to the model. In particular, an attacker can disrupt the integrity and availability of a model by simply reordering training batches, with no knowledge about either the model or the dataset. Indeed, the attacks presented here are not specific to the model or dataset, but rather target the stochastic nature of modern learning procedures. We extensively evaluate our attacks to find that the adversary can disrupt model training and even introduce backdoors. For integrity we find that the attacker can either stop the model from learning, or poison it to learn behaviours specified by the attacker. For availability we find that a single adversarially-ordered epoch can be enough to slow down model learning, or even to reset all of the learning progress. Such attacks have a long-term impact in that they decrease model performance hundreds of epochs after the attack took place. Reordering is a very powerful adversarial paradigm in that it removes the assumption that an adversary must inject adversarial data points or perturbations to perform training-time attacks. It reminds us that stochastic gradient descent relies on the assumption that data are sampled at random. If this randomness is compromised, then all bets are off.
Training machine learning (ML) models typically involves expensive iterative optimization. Once the model's final parameters are released, there is currently no mechanism for the entity which trained the model to prove that these parameters were indeed the result of this optimization procedure. Such a mechanism would support security of ML applications in several ways. For instance, it would simplify ownership resolution when multiple parties contest ownership of a specific model. It would also facilitate the distributed training across untrusted workers where Byzantine workers might otherwise mount a denial-of-service by returning incorrect model updates. In this paper, we remediate this problem by introducing the concept of proof-of-learning in ML. Inspired by research on both proof-of-work and verified computations, we observe how a seminal training algorithm, stochastic gradient descent, accumulates secret information due to its stochasticity. This produces a natural construction for a proof-of-learning which demonstrates that a party has expended the compute require to obtain a set of model parameters correctly. In particular, our analyses and experiments show that an adversary seeking to illegitimately manufacture a proof-of-learning needs to perform *at least* as much work than is needed for gradient descent itself. We also instantiate a concrete proof-of-learning mechanism in both of the scenarios described above. In model ownership resolution, it protects the intellectual property of models released publicly. In distributed training, it preserves availability of the training procedure. Our empirical evaluation validates that our proof-of-learning mechanism is robust to variance induced by the hardware (ML accelerators) and software stacks.
Machine learning benefits from large training datasets, which may not always be possible to collect by any single entity, especially when using privacy-sensitive data. In many contexts, such as healthcare and finance, separate parties may wish to collaborate and learn from each other's data but are prevented from doing so due to privacy regulations. Some regulations prevent explicit sharing of data between parties by joining datasets in a central location (confidentiality). Others also limit implicit sharing of data, e.g., through model predictions (privacy). There is currently no method that enables machine learning in such a setting, where both confidentiality and privacy need to be preserved, to prevent both explicit and implicit sharing of data. Federated learning only provides confidentiality, not privacy, since gradients shared still contain private information. Differentially private learning assumes unreasonably large datasets. Furthermore, both of these learning paradigms produce a central model whose architecture was previously agreed upon by all parties rather than enabling collaborative learning where each party learns and improves their own local model. We introduce Confidential and Private Collaborative (CaPC) learning, the first method provably achieving both confidentiality and privacy in a collaborative setting. We leverage secure multi-party computation (MPC), homomorphic encryption (HE), and other techniques in combination with privately aggregated teacher models. We demonstrate how CaPC allows participants to collaborate without having to explicitly join their training sets or train a central model. Each party is able to improve the accuracy and fairness of their model, even in settings where each party has a model that performs well on their own dataset or when datasets are not IID and model architectures are heterogeneous across parties.
Differentially private (DP) machine learning allows us to train models on private data while limiting data leakage. DP formalizes this data leakage through a cryptographic game, where an adversary must predict if a model was trained on a dataset D, or a dataset D' that differs in just one example.If observing the training algorithm does not meaningfully increase the adversary's odds of successfully guessing which dataset the model was trained on, then the algorithm is said to be differentially private. Hence, the purpose of privacy analysis is to upper bound the probability that any adversary could successfully guess which dataset the model was trained on.In our paper, we instantiate this hypothetical adversary in order to establish lower bounds on the probability that this distinguishing game can be won. We use this adversary to evaluate the importance of the adversary capabilities allowed in the privacy analysis of DP training algorithms.For DP-SGD, the most common method for training neural networks with differential privacy, our lower bounds are tight and match the theoretical upper bound. This implies that in order to prove better upper bounds, it will be necessary to make use of additional assumptions. Fortunately, we find that our attacks are significantly weaker when additional (realistic)restrictions are put in place on the adversary's capabilities.Thus, in the practical setting common to many real-world deployments, there is a gap between our lower bounds and the upper bounds provided by the analysis: differential privacy is conservative and adversaries may not be able to leak as much information as suggested by the theoretical bound.
Field-programmable gate arrays (FPGAs) are becoming widely used accelerators for a myriad of datacenter applications due to their flexibility and energy efficiency. Among these applications, FPGAs have shown promising results in accelerating low-latency real-time deep learning (DL) inference, which is becoming an indispensable component of many end-user applications. With the emerging research direction towards virtualized cloud FPGAs that can be shared by multiple users, the security aspect of FPGA-based DL accelerators requires careful consideration. In this work, we evaluate the security of DL accelerators against voltage-based integrity attacks in a multitenant FPGA scenario. We first demonstrate the feasibility of such attacks on a state-of-the-art Stratix 10 card using different attacker circuits that are logically and physically isolated in a separate attacker role, and cannot be flagged as malicious circuits by conventional bitstream checkers. We show that aggressive clock gating, an effective power-saving technique, can also be a potential security threat in modern FPGAs. Then, we carry out the attack on a DL accelerator running ImageNet classification in the victim role to evaluate the inherent resilience of DL models against timing faults induced by the adversary. We find that even when using the strongest attacker circuit, the prediction accuracy of the DL accelerator is not compromised when running at its safe operating frequency. Furthermore, we can achieve 1.18-1.31x higher inference performance by over-clocking the DL accelerator without affecting its prediction accuracy.
Current model extraction attacks assume that the adversary has access to a surrogate dataset with characteristics similar to the proprietary data used to train the victim model. This requirement precludes the use of existing model extraction techniques on valuable models, such as those trained on rare or hard to acquire datasets. In contrast, we propose data-free model extraction methods that do not require a surrogate dataset. Our approach adapts techniques from the area of data-free knowledge transfer for model extraction. As part of our study, we identify that the choice of loss is critical to ensuring that the extracted model is an accurate replica of the victim model. Furthermore, we address difficulties arising from the adversary's limited access to the victim model in a black-box setting. For example, we recover the model's logits from its probability predictions to approximate gradients. We find that the proposed data-free model extraction approach achieves high-accuracy with reasonable query complexity -- 0.99x and 0.92x the victim model accuracy on SVHN and CIFAR-10 datasets given 2M and 20M queries respectively.
Machine learning algorithms have been shown to be vulnerable to adversarial manipulation through systematic modification of inputs (e.g., adversarial examples) in domains such as image recognition. Under the default threat model, the adversary exploits the unconstrained nature of images; each feature (pixel) is fully under control of the adversary. However, it is not clear how these attacks translate to constrained domains that limit which and how features can be modified by the adversary (e.g., network intrusion detection). In this paper, we explore whether constrained domains are less vulnerable than unconstrained domains to adversarial example generation algorithms. We create an algorithm for generating adversarial sketches: targeted universal perturbation vectors which encode feature saliency within the envelope of domain constraints. To assess how these algorithms perform, we evaluate them in constrained (e.g., network intrusion detection) and unconstrained (e.g., image recognition) domains. The results demonstrate that our approaches generate misclassification rates in constrained domains that were comparable to those of unconstrained domains (greater than 95%). Our investigation shows that the narrow attack surface exposed by constrained domains is still sufficiently large to craft successful adversarial examples; and thus, constraints do not appear to make a domain robust. Indeed, with as little as five randomly selected features, one can still generate adversarial examples.