Latent Diffusion Models (LDMs) have achieved remarkable success in image generation tasks, yet their low barrier to customization poses severe threats related to art plagiarism. As a countermeasure, adversarial methods have been proposed to protect artworks from plagiarism. However, current methods suffer from limited effectiveness, high cost, and complex optimization. Moreover, their exploration and exploitation of LDM vulnerabilities remain limited, restricting effectiveness and applicability. To address this issue, we analyze the VAE and U-Net components of LDMs, revealing their vulnerabilities. Specifically, we study the response of U-Net to specific structural and frequency patterns in the latent space and find that it is susceptible to high-frequency and periodic latent features. Furthermore, we observe channel correlations during the VAE encoding process. Inspired by these, we propose QRShield, an efficient protection method that exploits the vulnerabilities of LDMs. By constructing high-frequency and periodic features consistent across latent channels and combining them with a momentum-based translation-invariant attack strategy, QRShield achieves stronger and more efficient protection. QRShield significantly improves protection performance in various fine-tuning settings, with over 10% gains in multiple metrics, a threefold increase in generation speed, and nearly 50% reduction in memory usage. Therefore, our work offers a more practical method to prevent AI art plagiarism.

Vision-language models (VLMs) have achieved remarkable success in various vision-language tasks, such as image captioning and visual question answering. However, these models often lack physical common sense, frequently failing to identify visually evident violations of common physical principles. Therefore, evaluating the VLMs' understanding of physical common sense is essential, which has not yet been systematically explored in existing research. To fill this gap, we introduce PhyVIB (Physical Common Sense Violation Image Benchmark). This novel benchmark consists of 16,000 images across eight categories, aiming to systematically assess the VLMs' capability to detect violations of physical common sense in images. Our evaluations show that even the state-of-the-art VLMs perform poorly on PhyVIB, highlighting a significant area for improvement. In response, we propose PhyDetector, a two-stage fine-tuning framework to enhance the VLMs' capability to detect violations of physical common sense. The first stage involves supervised fine-tuning, which equips the VLM with essential concepts related to visual physical anomalies. The second stage utilizes group relative policy optimization to enhance the VLM's multimodal reasoning capability on physical plausibility. Experimental results show that the model fine-tuned with PhyDetector can significantly outperform the state-of-the-art VLMs in physical common sense understanding. Our artifacts are available at https://github.com/ZitongWang018/PhyVIB.

Imperceptibility is one of the crucial requirements for adversarial examples. Previous adversarial attacks on 3D point cloud recognition suffer from noticeable outliers, resulting in low imperceptibility. We think that the drawbacks can be alleviated via taking the local curvature of the point cloud into consideration. Existing approaches introduce the local geometry distance into the attack objective function. However, their definition of the local geometry distance neglects different perceptibility of distortions along different directions. In this paper, we aim to enhance the imperceptibility of adversarial attacks on 3D point cloud recognition by better preserving the local curvature of the original 3D point clouds. To this end, we propose the Curvature-Invariant Method (CIM), which directly regularizes the back-propagated gradient during the generation of adversarial point clouds based on two assumptions. Specifically, we first decompose the back-propagated gradients into the tangent plane and the normal direction. Then we directly reduce the gradient along the large curvature direction on the tangent plane and only keep the gradient along the negative normal direction. Comprehensive experimental comparisons confirm the superiority of our approach. Notably, our strategy can achieve 7.2% and 14.5% improvements in Hausdorff distance and Gaussian curvature measurements of the imperceptibility.

With the great achievement of vision transformers (ViTs), transformer-based approaches have become the new paradigm for solving various computer vision tasks. However, recent research shows that similar to convolutional neural networks (CNNs), ViTs are still vulnerable to adversarial attacks. To explore the shared deficiency of models with different structures, researchers begin to analyze the cross-structure adversarial transferability, which is still under-explored. Therefore, in this work, we focus on the ViT attacks to improve the cross-structure transferability between the transformer-based and convolution-based models. Previous studies fail to thoroughly investigate the influence of the components inside the ViT models on adversarial transferability, leading to inferior performance. To overcome the drawback, we launch a motivating study by linearly down-scaling the gradients of components inside the ViT models to analyze their influence on adversarial transferability. Based on the motivating study, we find that the gradient of the skip connection most influences transferability and believe that back-propagating gradients from deeper blocks can enhance transferability. Therefore, we propose the Virtual Dense Connection method (VDC). Specifically, without changing the forward pass, we first recompose the original network to add virtual dense connections. Then we back-propagate gradients of deeper Attention maps and Multi-layer Perceptron (MLP) blocks via virtual dense connections when generating adversarial samples. Extensive experiments confirm the superiority of our proposed method over the state-of-the-art baselines, with an 8.2% improvement in transferability between ViT models and a 7.2% improvement in cross-structure transferability from ViTs to CNNs.

Various transfer attack methods have been proposed to evaluate the robustness of deep neural networks (DNNs). Although manifesting remarkable performance in generating untargeted adversarial perturbations, existing proposals still fail to achieve high targeted transferability. In this work, we discover that the adversarial perturbations' over-fitting towards source models of mediocre generalization capability can hurt their targeted transferability. To address this issue, we focus on enhancing the source model's gener-alization capability to improve its ability to conduct trans-ferable targeted adversarial attacks. In pursuit of this goal, we propose a novel model self-enhancement method that in-corporates two major components: Sharpness-Aware Self-Distillation (SASD) and Weight Scaling (WS). Specifically, SASD distills a fine-tuned auxiliary model, which mirrors the source model's structure, into the source model while flattening the source model's loss landscape. WS obtains an approximate ensemble of numerous pruned models to per-form model augmentation, which can be conveniently syn-ergized with SASD to elevate the source model's generalization capability and thus improve the resultant targeted per-turbations' transferability. Extensive experiments corrobo-rate the effectiveness of the proposed method. Notably, under the black-box setting, our approach can outperform the state-of-the-art baselines by a significant margin of 12.2% on average in terms of the obtained targeted transferability. Code is available at https://github.com/g4alllf/SASD.

Despite the excellent performance, deep neural networks (DNNs) have been shown to be vulnerable to adversarial examples. Besides, these examples are often transferable among different models. In other words, the same adversarial example can fool multiple models with different architectures at the same time. Based on this property, many black-box transfer-based attack techniques have been developed. However, current transfer-based attacks generally focus on the cross-architecture setting, where the attacker has access to the training data of the target model, which is not guaranteed in realistic situations. In this paper, we design a Cross-Domain Transfer-Based Attack (CDTA), which works in the cross-domain scenario. In this setting, attackers have no information about the target model, such as its architecture and training data. Specifically, we propose a contrastive spectral training method to train a feature extractor on a source domain (e.g., ImageNet) and use it to craft adversarial examples on target domains (e.g., Oxford 102 Flower). Our method corrupts the semantic information of the benign image by scrambling the outputs of both the intermediate feature layers and the final layer of the feature extractor. We evaluate CDTA with 16 target deep models on four datasets with widely varying styles. The results confirm that, in terms of the attack success rate, our approach can consistently outperform the state-of-the-art baselines by an average of 11.45% across all target models. Our code is available at https://github.com/LiulietLee/CDTA.

Vision transformers (ViTs) have been successfully deployed in a variety of computer vision tasks, but they are still vulnerable to adversarial samples. Transfer-based attacks use a local model to generate adversarial samples and directly transfer them to attack a target black-box model. The high efficiency of transfer-based attacks makes it a severe security threat to ViT-based applications. Therefore, it is vital to design effective transfer-based attacks to identify the deficiencies of ViTs beforehand in security-sensitive scenarios. Existing efforts generally focus on regularizing the input gradients to stabilize the updated direction of adversarial samples. However, the variance of the back-propagated gradients in intermediate blocks of ViTs may still be large, which may make the generated adversarial samples focus on some model-specific features and get stuck in poor local optima. To overcome the shortcomings of existing approaches, we propose the Token Gradient Regularization (TGR) method. According to the structural characteristics of ViTs, TGR reduces the variance of the back-propagated gradient in each internal block of ViTs in a token-wise manner and utilizes the regularized gradient to generate adversarial samples. Extensive experiments on attacking both ViTs and CNNs confirm the superiority of our approach. Notably, compared to the state-of-the-art transfer-based at-tacks, our TGR offers a performance improvement of 8.8% on average.

Deep neural networks have achieved unprecedented success on diverse vision tasks. However, they are vulnerable to adversarial noise that is imperceptible to humans. This phenomenon negatively affects their deployment in real-world scenarios, especially security-related ones. To evaluate the robustness of a target model in practice, transfer-based attacks craft adversarial samples with a local model and have attracted increasing attention from researchers due to their high efficiency. The state-of-the-art transfer-based attacks are generally based on data augmentation, which typically augments multiple training images from a linear path when learning adversarial samples. However, such methods selected the image augmentation path heuristically and may augment images that are semantics-inconsistent with the target images, which harms the transferability of the generated adversarial samples. To overcome the pitfall, we propose the Path-Augmented Method (PAM). Specifically, PAM first constructs a candidate augmentation path pool. It then settles the employed augmentation paths during adversarial sample generation with greedy search. Furthermore, to avoid augmenting semantics-inconsistent images, we train a Semantics Predictor (SP) to constrain the length of the augmentation path. Extensive experiments confirm that PAM can achieve an improvement of over 4.8% on average compared with the state-of-the-art baselines in terms of the attack success rates.

The exponential growth of social media platforms such as Twitter and Facebook has revolutionized textual communication and textual content publication in human society. However, they have been increasingly exploited to propagate toxic content, such as hate speech, malicious advertisement, and pornography, which can lead to highly negative impacts (e.g., harmful effects on teen mental health). Researchers and practitioners have been enthusiastically developing and extensively deploying textual content moderation software to address this problem. However, we find that malicious users can evade moderation by changing only a few words in the toxic content. Moreover, modern content moderation software's performance against malicious inputs remains underexplored. To this end, we propose MTTM, a Metamorphic Testing framework for Textual content Moderation software. Specifically, we conduct a pilot study on 2, 000 text messages collected from real users and summarize eleven metamorphic relations across three perturbation levels: character, word, and sentence. MTTM employs these metamorphic relations on toxic textual contents to generate test cases, which are still toxic yet likely to evade moderation. In our evaluation, we employ MTTM to test three commercial textual content moderation software and two state-of-the-art moderation algorithms against three kinds of toxic content. The results show that MTTM achieves up to 83.9%, 51%, and 82.5% error finding rates (EFR) when testing commercial moderation software provided by Google, Baidu, and Huawei, respectively, and it obtains up to 91.2% EFR when testing the state-of-the-art algorithms from the academy. In addition, we leverage the test cases generated by MTTM to retrain the model we explored, which largely improves model robustness 0% ~ 5.9% EFR) while maintaining the accuracy on the original test set. A demo can be found in this link1.

Despite their stunning performance, developing deep learning models from scratch is a formidable task. Therefore, it popularizes Machine-Learning-as-a-Service (MLaaS), where general users can access the trained models of MLaaS providers via Application Programming Interfaces (APIs) on a pay-per-query basis. Unfortunately, the success of MLaaS is under threat from model extraction attacks, where attackers intend to extract a local model of equivalent functionality to the target MLaaS model. However, existing studies on model extraction of text analytics APIs frequently assume adversaries have strong knowledge about the victim model, like its architecture and parameters, which hardly holds in practice. Besides, since the attacker's and the victim's training data can be considerably discrepant, it is non-trivial to perform efficient model extraction. In this paper, to advance the understanding of such attacks, we propose a framework, PEEP, for practical and efficient model extraction of sentiment analysis APIs with only query access. Specifically, PEEP features a learning-based scheme, which employs out-of-domain public corpora and a novel query strategy to construct proxy training data for model extraction. Besides, PEEP introduces a greedy search algorithm to settle an appropriate architecture for the extracted model. We conducted extensive experiments with two victim models across three datasets and two real-life commercial sentiment analysis APIs. Experimental results corroborate that PEEP can consistently outperform the state-of-the-art baselines in terms of effectiveness and efficiency.

Language models are known to be vulnerable to textual adversarial attacks, which add human-imperceptible perturbations to the input to mislead DNNs. It is thus imperative to devise effective attack algorithms to identify the deficiencies of DNNs before real-world deployment. However, existing word-level attacks have two major deficiencies: (1) They may change the semantics of the original sentence. (2) The generated adversarial sample can appear unnatural to humans due to the introduction of out-of-domain substitute words. In this paper, to address such drawbacks, we propose a semantics- and domain-aware word-level attack method. Specifically, we greedily replace the important words in a sentence with the ones suggested by a language model. The language model is trained to be semantics- and domain-aware via contrastive learning and in-domain pre-training. Furthermore, to balance the quality of adversarial examples and the attack success rate, we propose an iterative updating framework to optimize the contrastive learning loss and the in-domain pre-training loss in circular order. Comprehensive experimental comparisons confirm the superiority of our approach. Notably, compared with state-of-the-art benchmarks, our strategy can achieve over 3\% improvement in attack success rates and 9.8\% improvement in the quality of adversarial examples.

The exponential growth of social media platforms, such as Facebook, Instagram, Youtube, and TikTok, has revolutionized communication and content publication in human society. Users on these platforms can publish multimedia content that delivers information via the combination of text, audio, images, and video. Meanwhile, the multimedia content release facility has been increasingly exploited to propagate toxic content, such as hate speech, malicious advertisement, and pornography. To this end, content moderation software has been widely deployed on these platforms to detect and blocks toxic content. However, due to the complexity of content moderation models and the difficulty of understanding information across multiple modalities, existing content moderation software can fail to detect toxic content, which often leads to extremely negative impacts (e.g., harmful effects on teen mental health). We introduce Semantic Fusion, a general, effective methodology for validating multimedia content moderation software. Our key idea is to fuse two or more existing single-modal inputs (e.g., a textual sentence and an image) into a new input that combines the semantics of its ancestors in a novel manner and has toxic nature by construction. This fused input is then used for validating multimedia content moderation software. We realized Semantic Fusion as DUO, a practical content moderation software testing tool. In our evaluation, we employ DUO to test five commercial content moderation software and two state-of-the-art models against three kinds of toxic contents. The results show that DUO achieves up to 100% error finding rate (EFR) when testing moderation software and it obtains up to 94.1% EFR when testing the state-of-the-art models. In addition, we leverage the test cases generated by DUO to retrain the two models we explored, which largely improves model robustness (2.5%∼5.7% EFR) while maintaining the accuracy on the original test set.

Deep neural networks (DNNs) are vulnerable to adversarial attacks, which lead to incorrect predictions. In black-box settings, transfer attacks can be conveniently used to generate adversarial examples. However, such examples tend to overfit the specific architecture and feature representations of the source model, resulting in poor attack performance against other target models. To overcome this drawback, we propose a novel model modification-based transfer attack: Blurred-Dilated method (BD) in this paper. In summary, BD works by reducing downsampling while introducing BlurPool and dilated convolutions in the source model. Then BD employs the modified source model to generate adversarial samples. We think that BD can more comprehensively preserve the feature information than the original source model. It thus enables more thorough destruction of the image features, which can improve the transferability of the generated adversarial samples. Extensive experiments on the ImageNet dataset show that adversarial examples generated by BD achieve significantly higher transferability than the state-of-the-art baselines. Besides, BD can be conveniently combined with existing black-box attack techniques to further improve their performance.

Deep neural networks (DNNs) are known to be vulnerable to adversarial examples. It is thus imperative to devise effective attack algorithms to identify the deficiencies of DNNs beforehand in security-sensitive applications. To efficiently tackle the black-box setting where the target model's particulars are unknown, feature-level transfer-based attacks propose to contaminate the intermediate feature outputs of local models, and then directly employ the crafted adversarial samples to attack the target model. Due to the transferability of features, feature-level attacks have shown promise in synthesizing more transferable adversarial samples. However, existing feature-level attacks generally employ inaccurate neuron importance estimations, which deteriorates their transferability. To overcome such pitfalls, in this paper, we propose the Neuron Attribution-based Attack (NAA), which conducts feature-level attacks with more accurate neuron importance estimations. Specifically, we first completely attribute a model's output to each neuron in a middle layer. We then derive an approximation scheme of neuron attribution to tremendously reduce the computation overhead. Finally, we weight neurons based on their attribution results and launch feature-level attacks. Extensive experiments confirm the superiority of our approach to the state-of-the-art benchmarks. Our code is available at: hups.//rgithub.com/jprhang1810/NAA.

Although deep neural networks (DNNs) have achieved tremendous performance in diverse vision challenges, they are surprisingly susceptible to adversarial examples, which are born of intentionally perturbing benign samples in a human-imperceptible fashion. It thus poses security concerns on the deployment of DNNs in practice, particularly in safety- and security-sensitive domains. To investigate the robustness of DNNs, transfer-based attacks have attracted a growing interest recently due to their high practical applicability, where attackers craft adversarial samples with local models and employ the resultant samples to attack a remote black-box model. However, existing transfer-based attacks frequently suffer from low success rates due to overfitting to the adopted local model. To boost the transferability of adversarial samples, we propose to improve the robustness of synthesized adversarial samples via adversarial transformations. Specifically, we employ an adversarial transformation network to model the most harmful distortions that can destroy adversarial noises and require the synthesized adversarial samples to become resistant to such adversarial transformations. Extensive experiments on the ImageNet benchmark showcase the superiority of our method to state-of-the-art baselines in attacking both undefended and defended models.

The widespread deployment of deep models necessitates the assessment of model vulnerability in practice, especially for safety- and security-sensitive domains such as autonomous driving and medical diagnosis. Transfer-based attacks against image classifiers thus elicit mounting interest, where attackers are required to craft adversarial images based on local proxy models without the feedback information from remote target ones. However, under such a challenging but practical setup, the synthesized adversarial samples often achieve limited success due to overfitting to the local model employed. In this work, we propose a novel mechanism to alleviate the overfitting issue. It computes model attention over extracted features to regularize the search of adversarial examples, which prioritizes the corruption of critical features that are likely to be adopted by diverse architectures. Consequently, it can promote the transferability of resultant adversarial instances. Extensive experiments on ImageNet classifiers confirm the effectiveness of our strategy and its superiority to state-of-the-art benchmarks in both white-box and black-box settings.

With the growing prevalence of convolutional neural networks (CNNs), there is an urgent demand to explain their behaviors. Global explanations contribute to understanding model predictions on a whole category of samples, and thus have attracted increasing interest recently. However, existing methods overwhelmingly conduct separate input attribution or rely on local approximations of models, making them fail to offer faithful global explanations of CNNs. To overcome such drawbacks, we propose a novel two-stage framework, Attacking for Interpretability (AfI), which explains model decisions in terms of the importance of user-defined concepts. AfI first conducts a feature occlusion analysis, which resembles a process of attacking models to derive the category-wide importance of different features. We then map the feature importance to concept importance through ad-hoc semantic tasks. Experimental results confirm the effectiveness of AfI and its superiority in providing more accurate estimations of concept importance than existing proposals.

The exceptional performance of Deep neural networks (DNNs) encourages their deployment in safety-and dependability-critical systems. However, DNNs often demonstrate erroneous behaviors in real-world corner cases. Existing countermeasures center on improving the testing and bug-fixing practice. Unfortunately, building a bug-free DNN-based system is almost impossible currently due to its black-box nature, so anomaly detection is imperative in practice. Motivated by the idea of data validation in a traditional program, we propose and implement Deep Validation, a novel framework for detecting real-world error-inducing corner cases in a DNN-based system during runtime. We model the specifications of DNNs by resorting to their training data and cast checking input validity of DNNs as the problem of discrepancy estimation. Deep Validation achieves excellent detection results against various corner case scenarios across three popular datasets. Consequently, Deep Validation greatly complements existing efforts and is a crucial step toward building safe and dependable DNN-based systems.