Discretization Based Solutions for Secure Machine Learning Against Adversarial Attacks

Although deep neural networks (DNNs) have achieved high performance across various applications, they are often deceived by adversarial examples generated by adding small perturbations. To combat adversarial attacks, many detection methods have been proposed, including feature squeezing and trapdoor. However, these methods rely on the output of DNNs or involve training a separate network to detect adversarial examples, which leads to high computational costs and low efficiency. In this study, we propose a simple and effective approach called the entropy-based detector (EBD) to protect DNNs from various adversarial attacks. EBD detects adversarial examples by comparing the difference in entropy between the input sample before and after bit depth reduction. We show that EBD can detect over 98% of the adversarial examples generated by attacks using fast-gradient sign method, basic iterative method, momentum iterative method, DeepFool and CW attacks when the false positive rate is 2.5% for CIFAR-10 and ImageNet datasets.

Deep neural networks are vulnerable to maliciously crafted inputs called adversarial examples. Research on unprecedented adversarial attacks is significant since it can help strengthen the reliability of neural networks by alarming potential threats against them. However, since existing adversarial attacks disturb models unconditionally, the resulting adversarial examples increase their detectability through statistical observations or human inspection. To tackle this limitation, we propose hidden conditional adversarial attacks whose resultant adversarial examples disturb models only if the input images satisfy attackers’ pre-defined conditions. These hidden conditional adversarial examples have better stealthiness and controllability of their attack ability. Our experimental results on the CIFAR-10 and ImageNet datasets show their effectiveness and raise a serious concern about the vulnerability of CNNs against the novel attacks.

The Deep Neural Networks models have been extensively applied to many real-world areas due to their outstanding performance, especially to safety-critical applications such as autonomous vehicles, healthcare and face recognition. However, many types of research showed that they are easily disturbed by some carefully crafted tiny perturbations, which are called adversarial examples. Therefore, in this paper, a novel method is proposed as image preprocessing. When preprocessing, without discriminating the input images clean images or adversarial examples, the adaptive parameters adjust the images of label loss and pixel-level loss to jointly guide the model to generate reconstructed images and eliminate adversarial perturbation without affecting the images quality and classification result. Moreover, the model uses an architecture of a sparsity constraint that ensures that a neuron fires only for meaningful patterns, limiting the influence of adversarial perturbation on clean images. The defense proposed against attacks of FGSM, MI-FGSM, C&W, PGD, RAND+FGSM under the MNIST and CIFAR-10 datasets. The experimental results show that the target model can reach the average accuracy of 98% and 92% separately when the MNIST and CIFAR-10 datasets are attacked through the defense proposed, demonstrating the effectiveness of the proposed method.

The rapid development of deep neural networks (DNN) has promoted the widespread application of image recognition, natural language processing, and autonomous driving. However, DNN is vulnerable to adversarial examples, such as an input sample with imperceptible perturbation which can easily invalidate the DNN and even deliberately modify the classification results. Therefore, this article proposes a preprocessing defense framework based on image compression reconstruction to achieve adversarial example defense. Firstly, the defense framework performs pixel depth compression on the input image based on the sensitivity of the adversarial example to eliminate adversarial perturbations. Secondly, we use the super-resolution image reconstruction network to restore the image quality and then map the adversarial example to the clean image. Therefore, there is no need to modify the network structure of the classifier model, and it can be easily combined with other defense methods. Finally, we evaluate the algorithm with MNIST, Fashion-MNIST, and CIFAR-10 datasets; the experimental results show that our approach outperforms current techniques in the task of defending against adversarial example attacks.

We find that images contain intrinsic structure that enables the reversal of many adversarial attacks. Attack vectors cause not only image classifiers to fail, but also collaterally disrupt incidental structure in the image. We demonstrate that modifying the attacked image to restore the natural structure will reverse many types of attacks, providing a defense. Experiments demonstrate significantly improved robustness for several state-of-the-art models across the CIFAR-10, CIFAR-100, SVHN, and ImageNet datasets. Our results show that our defense is still effective even if the attacker is aware of the defense mechanism. Since our defense is deployed during inference instead of training, it is compatible with pre-trained networks as well as most other defenses. Our results suggest deep networks are vulnerable to adversarial examples partly because their representations do not enforce the natural structure of images.

Some recent articles have revealed that synthetic aperture radar automatic target recognition (SAR-ATR) models based on deep learning are vulnerable to the attacks of adversarial examples and cause security problems. The adversarial attack can make a deep convolutional neural network (CNN)-based SAR-ATR system output the intended wrong label predictions by adding small adversarial perturbations to the SAR images. The existing optimization-based adversarial attack methods generate adversarial examples by minimizing the mean-squared reconstruction error, causing smooth target edge and blurry weak scattering centers in SAR images. In this paper, we build a UNet-generative adversarial network (GAN) to refine the generation of the SAR-ATR models’ adversarial examples. The UNet learns the separable features of the targets and generates the adversarial examples of SAR images. The GAN makes the generated adversarial examples approximate to real SAR images (with sharp target edge and explicit weak scattering centers) and improves the generation efficiency. We carry out abundant experiments using the proposed adversarial attack algorithm to fool the SAR-ATR models based on several advanced CNNs, which are trained on the measured SAR images of the ground vehicle targets. The quantitative and qualitative results demonstrate the high-quality adversarial example generation and excellent attack effectiveness and efficiency improvement.

Deep learning has revolutionized image recognition. One significant obstacle still remains, the vulnerability of these models to adversarial attacks. These attacks manipulate images with subtle changes that cause CNN misclassification. While methods, such as adversarial training, have been proposed to defend against adversarial attacks, they incur additional training costs, either extra input samples or auxiliary models. In this work, we propose an efficient approach to deploying robust models that utilize image transformations to remove adversarial noises. We investigate the performance of simple transformations and report several effective ones, including Affine blur, Gaussian blur, Median blur, and Bilateral blur against various adversarial attack methods, such as Fast Gradient Sign Method (FGSM), Randomized + FGSM (RFGSM), and Projected Gradient Descent (PGD). We apply these image transformation techniques to the widely used ImageNet dataset, and experimental results demonstrate the potential of image transformation methods as a strong defense against adversarial attacks in deep learning-based image classification systems, especially when combined with cutting-edge neural network architectures such as ResNet50 and DenseNet121. Our comprehensive results show that these transformations can significantly improve the robustness of CNN models against adversarial attacks on ImageNet, achieving a recovery rate of up to 85% to 90% without incurring extra resource costs.

Language Models today provide a high accuracy across a large number of\ndownstream tasks. However, they remain susceptible to adversarial attacks,\nparticularly against those where the adversarial examples maintain considerable\nsimilarity to the original text. Given the multilingual nature of text, the\neffectiveness of adversarial examples across translations and how machine\ntranslations can improve the robustness of adversarial examples remain largely\nunexplored. In this paper, we present a comprehensive study on the robustness\nof current text adversarial attacks to round-trip translation. We demonstrate\nthat 6 state-of-the-art text-based adversarial attacks do not maintain their\nefficacy after round-trip translation. Furthermore, we introduce an\nintervention-based solution to this problem, by integrating Machine Translation\ninto the process of adversarial example generation and demonstrating increased\nrobustness to round-trip translation. Our results indicate that finding\nadversarial examples robust to translation can help identify the insufficiency\nof language models that is common across languages, and motivate further\nresearch into multilingual adversarial attacks.\n

State-of-the-art neural network models are actively used in various fields, but it is well-known that they are vulnerable to adversarial example attacks. Throughout the efforts to make the models robust against adversarial example attacks, it has been found to be a very difficult task. While many defense approaches were shown to be not effective, adversarial training remains as one of the promising methods. In adversarial training, the training data are augmented by “adversarial” samples generated using an attack algorithm. If the attacker uses a similar attack algorithm to generate adversarial examples, the adversarially trained network can be quite robust to the attack. However, there are numerous ways of creating adversarial examples, and the defender does not know what algorithm the attacker may use. A natural question is: Can we use adversarial training to train a model robust to multiple types of attack? Previous work have shown that, when a network is trained with adversarial examples generated from multiple attack methods, the network is still vulnerable to white-box attacks where the attacker has complete access to the model parameters. In this paper, we study this question in the context of black-box attacks, which can be a more realistic assumption for practical applications. Experiments with the MNIST dataset show that adversarially training a network with an attack method helps defending against that particular attack method, but has limited effect for other attack methods. In addition, even if the defender trains a network with multiple types of adversarial examples and the attacker attacks with one of the methods, the network could lose accuracy to the attack if the attacker uses a different data augmentation strategy on the target network. These results show that it is very difficult to make a robust network using adversarial training, even for black-box settings where the attacker has restricted information on the target network.

Adversarial examples: A survey of attacks and defenses in deep learning-enabled cybersecurity systems

Recent studies have shown that deep learning models such as image classification and object detection models are vulnerable to attacks from adversarial examples. These malicious examples may trigger potential security threats and cause damage in security-related fields. In this paper, in order to find model robustness weakness, we obtain the adversarial robustness metric against several adversarial attacks, including gradient-based attacks, optimization-based attacks, etc. The metric quantizes the adversarial robustness of deep learning models and can be used as a guide for extra model training. Meanwhile, we design a testing platform to achieve model adversarial robustness evaluation. Utilizing the front and back end separation strategy, the testing platform consist of 2 modules: one is multiple adversarial example attacks executor and the other is HCI module. In addition, the tool allows parallel comparison among different adversarial attack algorithms under given conditions and measures the quality of generated adversarial examples utilizing FID(Fréchet Inception Distance) metrics. Besides, we developed ERFGSM(edge-RFGSM), a new gradient-based attack using edge information through canny operator.

Deep neural networks (DNNs) have gained widespread adoption in computer vision. Unfortunately, state-of-the-art DNNs are vulnerable to adversarial example (AE) attacks, where an adversary introduces imperceptible perturbations to a test example for defrauding DNNs. The obstacles have urged intensive research on improving the DNN robustness via adversarial training, that is, the clean data set is blended with adversarial examples to carry out training. However, the adversarial example attack technologies are open-ended, and the adversarial training is insufficient to focus on improving robustness performance. To circumvent this limitation, we mitigate adversarial example attacks from another perspective, which aims at detecting adversarial examples. Feature autoencoder detector (FADetector), a novel defense framework that exploits feature knowledge is proposed. One of the hallmarks of FADetector is to not involve adversarial examples to train the detector. Our extensive evaluation on MNIST and CIFAR-10 data sets demonstrates that our defense outperforms the conventional autoencoder detectors in terms of detection accuracy.

Most deep neural networks are vulnerable to adversarial attacks, making it crucial to develop effective defenses against them. Adversarial examples are created by applying slight perturbations either to the entire image or specific regions, depending on attack methods. By contrast, benign examples typically lead models to focus attention on contiguous regions of the image during classification. Therefore, adversarial examples can be detected effectively by leveraging attention values. First, we conduct simple experiment to investigate how the model’s attention responds to adversarial perturbations. Second, we derive features by calculating the changes in attention values across ViT layers for adversarial and benign examples. These attention-derived features are then fed into the detection model, ResNet-50, to learn the differences in the activation locations between both examples. Third, we demonstrate the effectiveness of our approach by thorough experiments on various attack methods, datasets, and target models. The experiments prove the effectiveness of our approach, achieving superior performance compared to baseline methods on the ImageNet dataset across ViT-B and ViT-L. Our findings underscore the potential of utilizing attention-derived features as powerful indicators for enhancing the resilience of ViT models against adversarial perturbations.

Recent research shows that deep neural networks make wrong predictions when faced with adversarial examples with small perturbations added. In the setting of white-box attack, it is easy to generate adversarial samples with high attack success rate through gradients. But in reality, gradients are usually unavailable. At present, most black-box attack achieves the purpose of the attack by adding small perturbations, the intensity of the perturbation and the attack success rate are a kind of trade off. However, the noise added by most methods is meaningless. This paper introduces a novel adversarial attack algorithm called Pixel Logo Attack (PLA), which rationalizes noise by arranging pixel patterns into a logo-style, thereby presenting itself as an identity for protecting copyright information. Unlike most existing methods, this method can completely expose the added noise to the user without arousing user suspicion, and does not affect the usage of image. We use the differential evolution(DE) to search for suitable pixel positions and RGB values, and compare the performance of PLA with the state-of-the-art adversarial attack algorithms on the CIFAR-10 and ImageNet datasets. Experimental results show that PLA has good performance in solving black-box adversarial attack problems, especially non-targeted attack.

Deep neural network-based synthetic aperture radar (SAR) automatic target recognition (ATR) systems are susceptible to attack by adversarial examples, which leads to misclassification by the SAR ATR system, resulting in theoretical model robustness problems and security problems in practice. Inspired by optical images, current SAR ATR adversarial example generation is performed in the image domain. However, the imaging principle of SAR images is based on the imaging of the echo signals interacting between the SAR and objects. Generating adversarial examples only in the image domain cannot change the physical world to achieve adversarial attacks. To solve these problems, this article proposes a framework for generating SAR adversarial examples in a 3D physical scene. First, adversarial attacks are implemented in the 2D image space, and the perturbation in the image space is converted into simulated rays that constitute SAR images through backpropagation optimization methods. The mapping between the simulated rays constituting SAR images and the 3D model is established through coordinate transformation, and point correspondence to triangular faces and intensity values to texture parameters are established. Thus, the simulated rays constituting SAR images are mapped to the 3D model, and the perturbation in the 2D image space is converted back to the 3D physical space to obtain the position and intensity of the perturbation in the 3D physical space, thereby achieving physical adversarial attacks. The experimental results show that our attack method can effectively perform SAR adversarial attacks in the physical world. In the digital world, we achieved an average fooling rate of up to 99.02% for three objects in six classification networks. In the physical world, we achieved an average fooling rate of up to 97.87% for these objects, with a certain degree of transferability across the six different network architectures. To the best of our knowledge, this is the first work to implement physical attacks in a full physical simulation condition. Our research establishes a theoretical foundation for the future concealment of SAR targets in practical settings and offers valuable insights for enhancing the attack and defense capabilities of subsequent DNNs in SAR ATR systems.