2023-12-13 更新
On the Robustness of Large Multimodal Models Against Image Adversarial Attacks
Authors:Xuanming Cui, Alejandro Aparcedo, Young Kyun Jang, Ser-Nam Lim
Recent advances in instruction tuning have led to the development of State-of-the-Art Large Multimodal Models (LMMs). Given the novelty of these models, the impact of visual adversarial attacks on LMMs has not been thoroughly examined. We conduct a comprehensive study of the robustness of various LMMs against different adversarial attacks, evaluated across tasks including image classification, image captioning, and Visual Question Answer (VQA). We find that in general LMMs are not robust to visual adversarial inputs. However, our findings suggest that context provided to the model via prompts, such as questions in a QA pair helps to mitigate the effects of visual adversarial inputs. Notably, the LMMs evaluated demonstrated remarkable resilience to such attacks on the ScienceQA task with only an 8.10% drop in performance compared to their visual counterparts which dropped 99.73%. We also propose a new approach to real-world image classification which we term query decomposition. By incorporating existence queries into our input prompt we observe diminished attack effectiveness and improvements in image classification accuracy. This research highlights a previously under-explored facet of LMM robustness and sets the stage for future work aimed at strengthening the resilience of multimodal systems in adversarial environments.
PDF
点此查看论文截图
Patch-MI: Enhancing Model Inversion Attacks via Patch-Based Reconstruction
Authors:Jonggyu Jang, Hyeonsu Lyu, Hyun Jong Yang
Model inversion (MI) attacks aim to reveal sensitive information in training datasets by solely accessing model weights. Generative MI attacks, a prominent strand in this field, utilize auxiliary datasets to recreate target data attributes, restricting the images to remain photo-realistic, but their success often depends on the similarity between auxiliary and target datasets. If the distributions are dissimilar, existing MI attack attempts frequently fail, yielding unrealistic or target-unrelated results. In response to these challenges, we introduce a groundbreaking approach named Patch-MI, inspired by jigsaw puzzle assembly. To this end, we build upon a new probabilistic interpretation of MI attacks, employing a generative adversarial network (GAN)-like framework with a patch-based discriminator. This approach allows the synthesis of images that are similar to the target dataset distribution, even in cases of dissimilar auxiliary dataset distribution. Moreover, we artfully employ a random transformation block, a sophisticated maneuver that crafts generalized images, thus enhancing the efficacy of the target classifier. Our numerical and graphical findings demonstrate that Patch-MI surpasses existing generative MI methods in terms of accuracy, marking significant advancements while preserving comparable statistical dataset quality. For reproducibility of our results, we make our source code publicly available in https://github.com/jonggyujang0123/Patch-Attack.
PDF 11 pages
点此查看论文截图
Divide-and-Conquer Attack: Harnessing the Power of LLM to Bypass the Censorship of Text-to-Image Generation Model
Authors:Yimo Deng, Huangxun Chen
Text-to-image generative models offer many innovative services but also raise ethical concerns due to their potential to generate unethical images. Most publicly available text-to-image models employ safety filters to prevent unintended generation intents. In this work, we introduce the Divide-and-Conquer Attack to circumvent the safety filters of state-of-the-art text-to-image models. Our attack leverages LLMs as agents for text transformation, creating adversarial prompts from sensitive ones. We have developed effective helper prompts that enable LLMs to break down sensitive drawing prompts into multiple harmless descriptions, allowing them to bypass safety filters while still generating sensitive images. This means that the latent harmful meaning only becomes apparent when all individual elements are drawn together. Our evaluation demonstrates that our attack successfully circumvents the closed-box safety filter of SOTA DALLE-3 integrated natively into ChatGPT to generate unethical images. This approach, which essentially uses LLM-generated adversarial prompts against GPT-4-assisted DALLE-3, is akin to using one’s own spear to breach their shield. It could have more severe security implications than previous manual crafting or iterative model querying methods, and we hope it stimulates more attention towards similar efforts. Our code and data are available at: https://github.com/researchcode001/Divide-and-Conquer-Attack
PDF 20 pages,6 figures, under review
点此查看论文截图
SSTA: Salient Spatially Transformed Attack
Authors:Renyang Liu, Wei Zhou, Sixin Wu, Jun Zhao, Kwok-Yan Lam
Extensive studies have demonstrated that deep neural networks (DNNs) are vulnerable to adversarial attacks, which brings a huge security risk to the further application of DNNs, especially for the AI models developed in the real world. Despite the significant progress that has been made recently, existing attack methods still suffer from the unsatisfactory performance of escaping from being detected by naked human eyes due to the formulation of adversarial example (AE) heavily relying on a noise-adding manner. Such mentioned challenges will significantly increase the risk of exposure and result in an attack to be failed. Therefore, in this paper, we propose the Salient Spatially Transformed Attack (SSTA), a novel framework to craft imperceptible AEs, which enhance the stealthiness of AEs by estimating a smooth spatial transform metric on a most critical area to generate AEs instead of adding external noise to the whole image. Compared to state-of-the-art baselines, extensive experiments indicated that SSTA could effectively improve the imperceptibility of the AEs while maintaining a 100\% attack success rate.
PDF
点此查看论文截图