Mikhail Pautov

Andrei Chertkov, Olga Tsymboi, Mikhail Pautov et al.

Neural networks are deployed widely in natural language processing tasks on the industrial scale, and perhaps the most often they are used as compounds of automatic machine translation systems. In this work, we present a simple approach to fool state-of-the-art machine translation tools in the task of translation from Russian to English and vice versa. Using a novel black-box gradient-free tensor-based optimizer, we show that many online translation tools, such as Google, DeepL, and Yandex, may both produce wrong or offensive translations for nonsensical adversarial input queries and refuse to translate seemingly benign input phrases. This vulnerability may interfere with understanding a new language and simply worsen the user's experience while using machine translation systems, and, hence, additional improvements of these tools are required to establish better translation.

Dmitrii Korzh, Mikhail Pautov, Olga Tsymboi et al.

Randomized smoothing is the state-of-the-art approach to construct image classifiers that are provably robust against additive adversarial perturbations of bounded magnitude. However, it is more complicated to construct reasonable certificates against semantic transformation (e.g., image blurring, translation, gamma correction) and their compositions. In this work, we propose \emph{General Lipschitz (GL),} a new framework to certify neural networks against composable resolvable semantic perturbations. Within the framework, we analyze transformation-dependent Lipschitz-continuity of smoothed classifiers w.r.t. transformation parameters and derive corresponding robustness certificates. Our method performs comparably to state-of-the-art approaches on the ImageNet dataset.

Artem Dvirniak, Evgeny Kushnir, Dmitrii Tarasov et al.

The modern generative audio models can be used by an adversary in an unlawful manner, specifically, to impersonate other people to gain access to private information. To mitigate this issue, speech deepfake detection (SDD) methods started to evolve. Unfortunately, current SDD methods generally suffer from the lack of generalization to new audio domains and generators. More than that, they lack interpretability, especially human-like reasoning that would naturally explain the attribution of a given audio to the bona fide or spoof class and provide human-perceptible cues. In this paper, we propose HIR-SDD, a novel SDD framework that combines the strengths of Large Audio Language Models (LALMs) with the chain-of-thought reasoning derived from the novel proposed human-annotated dataset. Experimental evaluation demonstrates both the effectiveness of the proposed method and its ability to provide reasonable justifications for predictions.

Evgeny Kushnir, Alexandr Kozodaev, Dmitrii Korzh et al.

Recent advances in generative models have amplified the risk of malicious misuse of speech synthesis technologies, enabling adversaries to impersonate target speakers and access sensitive resources. Although speech deepfake detection has progressed rapidly, most existing countermeasures lack formal robustness guarantees or fail to generalize to unseen generation techniques. We propose PV-VASM, a probabilistic framework for verifying the robustness of voice anti-spoofing models (VASMs). PV-VASM estimates the probability of misclassification under text-to-speech (TTS), voice cloning (VC), and parametric signal transformations. The approach is model-agnostic and enables robustness verification against unseen speech synthesis techniques and input perturbations. We derive a theoretical upper bound on the error probability and validate the method across diverse experimental settings, demonstrating its effectiveness as a practical robustness verification tool.

Anna Chistyakova, Mikhail Pautov

Black-box adversarial attacks are widely used as tools to test the robustness of deep neural networks against malicious perturbations of input data aimed at a specific change in the output of the model. Such methods, although they remain empirically effective, usually do not guarantee that an adversarial example can be found for a particular model. In this paper, we propose Contract And Conquer (CAC), an approach to provably compute adversarial examples for neural networks in a black-box manner. The method is based on knowledge distillation of a black-box model on an expanding distillation dataset and precise contraction of the adversarial example search space. CAC is supported by the transferability guarantee: we prove that the method yields an adversarial example for the black-box model within a fixed number of algorithm iterations. Experimentally, we demonstrate that the proposed approach outperforms existing state-of-the-art black-box attack methods on ImageNet dataset for different target models, including vision transformers.

Anna Chistyakova, Mikhail Pautov

Being trained on large and diverse datasets, visual foundation models (VFMs) can be fine-tuned to achieve remarkable performance and efficiency in various downstream computer vision tasks. The high computational cost of data collection and training makes these models valuable assets, which motivates some VFM owners to distribute them alongside a license to protect their intellectual property rights. In this paper, we propose an approach to ownership verification of visual foundation models that leverages a small encoder-decoder network to embed digital watermarks into an internal representation of a hold-out set of input images. The method is based on random watermark embedding, which makes the watermark statistics detectable in functional copies of the watermarked model. Both theoretically and experimentally, we demonstrate that the proposed method yields a low probability of false detection for non-watermarked models and a low probability of false misdetection for watermarked models.

Andrey V. Galichin, Mikhail Pautov, Alexey Zhavoronkin et al.

While Deep Neural Networks (DNNs) have demonstrated remarkable performance in tasks related to perception and control, there are still several unresolved concerns regarding the privacy of their training data, particularly in the context of vulnerability to Membership Inference Attacks (MIAs). In this paper, we explore a connection between the susceptibility to membership inference attacks and the vulnerability to distillation-based functionality stealing attacks. In particular, we propose {GLiRA}, a distillation-guided approach to membership inference attack on the black-box neural network. We observe that the knowledge distillation significantly improves the efficiency of likelihood ratio of membership inference attack, especially in the black-box setting, i.e., when the architecture of the target model is unknown to the attacker. We evaluate the proposed method across multiple image classification datasets and models and demonstrate that likelihood ratio attacks when guided by the knowledge distillation, outperform the current state-of-the-art membership inference attacks in the black-box setting.

Dmitrii Korzh, Elvir Karimov, Mikhail Pautov et al.

Speaker recognition technology is applied to various tasks, from personal virtual assistants to secure access systems. However, the robustness of these systems against adversarial attacks, particularly to additive perturbations, remains a significant challenge. In this paper, we pioneer applying robustness certification techniques to speaker recognition, initially developed for the image domain. Our work covers this gap by transferring and improving randomized smoothing certification techniques against norm-bounded additive perturbations for classification and few-shot learning tasks to speaker recognition. We demonstrate the effectiveness of these methods on VoxCeleb 1 and 2 datasets for several models. We expect this work to improve the robustness of voice biometrics and accelerate the research of certification methods in the audio domain.

Kirill Lukyanov, Andrew Perminov, Denis Turdakov et al.

The vulnerability of artificial neural networks to adversarial perturbations in the black-box setting is widely studied in the literature. The majority of attack methods to construct these perturbations suffer from an impractically large number of queries required to find an adversarial example. In this work, we focus on knowledge distillation as an approach to conduct transfer-based black-box adversarial attacks and propose an iterative training of the surrogate model on an expanding dataset. This work is the first, to our knowledge, to provide provable guarantees on the success of knowledge distillation-based attack on classification neural networks: we prove that if the student model has enough learning capabilities, the attack on the teacher model is guaranteed to be found within the finite number of distillation iterations.

Richik Sengupta, Mikhail Pautov

The following hypothesis was formulated by Goreinov, Tyrtyshnikov, and Zamarashkin in \cite{goreinov1997theory}. If $U$ is $n\times k$ real matrix with the orthonormal columns $(n>k)$, then there exists a submatrix $Q$ of $U$ of size $k\times k$ such that its smallest singular value is at least $\frac{1}{\sqrt{n}}.$ Although this statement is supported by numerical experiments, the problem remains open for all $1<k<n-1,$ except for the case of $n \leq 4,\ k=2.$ In this work, we provide a proof for the case $k=2.$

Anna Chistyakova, Mikhail Pautov

Being trained on large and vast datasets, visual foundation models (VFMs) can be fine-tuned for diverse downstream tasks, achieving remarkable performance and efficiency in various computer vision applications. The high computation cost of data collection and training motivates the owners of some VFMs to distribute them alongside the license to protect their intellectual property rights. However, a dishonest user of the protected model's copy may illegally redistribute it, for example, to make a profit. As a consequence, the development of reliable ownership verification tools is of great importance today, since such methods can be used to differentiate between a redistributed copy of the protected model and an independent model. In this paper, we propose an approach to ownership verification of visual foundation models by fine-tuning a small set of expressive layers of a VFM along with a small encoder-decoder network to embed digital watermarks into an internal representation of a hold-out set of input images. Importantly, the watermarks embedded remain detectable in the functional copies of the protected model, obtained, for example, by fine-tuning the VFM for a particular downstream task. Theoretically and experimentally, we demonstrate that the proposed method yields a low probability of false detection of a non-watermarked model and a low probability of false misdetection of a watermarked model.

Mikhail Pautov, Danil Ivanov, Andrey V. Galichin et al.

Generative models that can produce realistic images have improved significantly in recent years. The quality of the generated content has increased drastically, so sometimes it is very difficult to distinguish between the real images and the generated ones. Such an improvement comes at a price of ethical concerns about the usage of the generative models: the users of generative models can improperly claim ownership of the generated content protected by a license. In this paper, we propose an approach to embed watermarks into the generated content to allow future detection of the generated content and identification of the user who generated it. The watermark is embedded during the inference of the model, so the proposed approach does not require the retraining of the latter. We prove that watermarks embedded are guaranteed to be robust against additive perturbations of a bounded magnitude. We apply our method to watermark diffusion models and show that it matches state-of-the-art watermarking schemes in terms of robustness to different types of synthetic watermark removal attacks.

Ekaterina Shumitskaya, Mikhail Pautov, Dmitriy Vatolin et al.

Most modern No-Reference Image-Quality Assessment (NR-IQA) metrics are based on neural networks vulnerable to adversarial attacks. Attacks on such metrics lead to incorrect image/video quality predictions, which poses significant risks, especially in public benchmarks. Developers of image processing algorithms may unfairly increase the score of a target IQA metric without improving the actual quality of the adversarial image. Although some empirical defenses for IQA metrics were proposed, they do not provide theoretical guarantees and may be vulnerable to adaptive attacks. This work focuses on developing a provably robust no-reference IQA metric. Our method is based on Median Smoothing (MS) combined with an additional convolution denoiser with ranking loss to improve the SROCC and PLCC scores of the defended IQA metric. Compared with two prior methods on three datasets, our method exhibited superior SROCC and PLCC scores while maintaining comparable certified guarantees.

Mikhail Pautov, Nikita Bogdanov, Stanislav Pyatkin et al.

As deep learning (DL) models are widely and effectively used in Machine Learning as a Service (MLaaS) platforms, there is a rapidly growing interest in DL watermarking techniques that can be used to confirm the ownership of a particular model. Unfortunately, these methods usually produce watermarks susceptible to model stealing attacks. In our research, we introduce a novel trigger set-based watermarking approach that demonstrates resilience against functionality stealing attacks, particularly those involving extraction and distillation. Our approach does not require additional model training and can be applied to any model architecture. The key idea of our method is to compute the trigger set, which is transferable between the source model and the set of proxy models with a high probability. In our experimental study, we show that if the probability of the set being transferable is reasonably high, it can be effectively used for ownership verification of the stolen model. We evaluate our method on multiple benchmarks and show that our approach outperforms current state-of-the-art watermarking techniques in all considered experimental setups.

Mikhail Pautov, Olesya Kuznetsova, Nurislam Tursynbek et al.

Randomized smoothing is considered to be the state-of-the-art provable defense against adversarial perturbations. However, it heavily exploits the fact that classifiers map input objects to class probabilities and do not focus on the ones that learn a metric space in which classification is performed by computing distances to embeddings of classes prototypes. In this work, we extend randomized smoothing to few-shot learning models that map inputs to normalized embeddings. We provide analysis of Lipschitz continuity of such models and derive robustness certificate against $\ell_2$-bounded perturbations that may be useful in few-shot learning scenarios. Our theoretical results are confirmed by experiments on different datasets.

Mikhail Pautov, Nurislam Tursynbek, Marina Munkhoeva et al.

In safety-critical machine learning applications, it is crucial to defend models against adversarial attacks -- small modifications of the input that change the predictions. Besides rigorously studied $\ell_p$-bounded additive perturbations, recently proposed semantic perturbations (e.g. rotation, translation) raise a serious concern on deploying ML systems in real-world. Therefore, it is important to provide provable guarantees for deep learning models against semantically meaningful input transformations. In this paper, we propose a new universal probabilistic certification approach based on Chernoff-Cramer bounds that can be used in general attack settings. We estimate the probability of a model to fail if the attack is sampled from a certain distribution. Our theoretical findings are supported by experimental results on different datasets.

Mikhail Pautov, Grigorii Melnikov, Edgar Kaziakhmedov et al.

Recent works showed the vulnerability of image classifiers to adversarial attacks in the digital domain. However, the majority of attacks involve adding small perturbation to an image to fool the classifier. Unfortunately, such procedures can not be used to conduct a real-world attack, where adding an adversarial attribute to the photo is a more practical approach. In this paper, we study the problem of real-world attacks on face recognition systems. We examine security of one of the best public face recognition systems, LResNet100E-IR with ArcFace loss, and propose a simple method to attack it in the physical world. The method suggests creating an adversarial patch that can be printed, added as a face attribute and photographed; the photo of a person with such attribute is then passed to the classifier such that the classifier's recognized class changes from correct to the desired one. Proposed generating procedure allows projecting adversarial patches not only on different areas of the face, such as nose or forehead but also on some wearable accessory, such as eyeglasses.

Edgar Kaziakhmedov, Klim Kireev, Grigorii Melnikov et al.

Recent studies proved that deep learning approaches achieve remarkable results on face detection task. On the other hand, the advances gave rise to a new problem associated with the security of the deep convolutional neural network models unveiling potential risks of DCNNs based applications. Even minor input changes in the digital domain can result in the network being fooled. It was shown then that some deep learning-based face detectors are prone to adversarial attacks not only in a digital domain but also in the real world. In the paper, we investigate the security of the well-known cascade CNN face detection system - MTCNN and introduce an easily reproducible and a robust way to attack it. We propose different face attributes printed on an ordinary white and black printer and attached either to the medical face mask or to the face directly. Our approach is capable of breaking the MTCNN detector in a real-world scenario.