VEAttack: Downstream-agnostic Vision Encoder Attack against Large Vision Language Models
This addresses robustness concerns for LVLMs used in diverse multimodal applications, though it is incremental as it builds on existing attack methods by focusing on the vision encoder.
The paper tackles the vulnerability of Large Vision-Language Models (LVLMs) to adversarial attacks by proposing VEAttack, a method that targets only the vision encoder to reduce computational costs and task dependence, achieving performance degradation of 94.5% on image caption and 75.7% on visual question answering tasks.
Large Vision-Language Models (LVLMs) have demonstrated remarkable capabilities in multimodal understanding and generation, yet their vulnerability to adversarial attacks raises significant robustness concerns. While existing effective attacks always focus on task-specific white-box settings, these approaches are limited in the context of LVLMs, which are designed for diverse downstream tasks and require expensive full-model gradient computations. Motivated by the pivotal role and wide adoption of the vision encoder in LVLMs, we propose a simple yet effective Vision Encoder Attack (VEAttack), which targets the vision encoder of LVLMs only. Specifically, we propose to generate adversarial examples by minimizing the cosine similarity between the clean and perturbed visual features, without accessing the following large language models, task information, and labels. It significantly reduces the computational overhead while eliminating the task and label dependence of traditional white-box attacks in LVLMs. To make this simple attack effective, we propose to perturb images by optimizing image tokens instead of the classification token. We provide both empirical and theoretical evidence that VEAttack can easily generalize to various tasks. VEAttack has achieved a performance degradation of 94.5% on image caption task and 75.7% on visual question answering task. We also reveal some key observations to provide insights into LVLM attack/defense: 1) hidden layer variations of LLM, 2) token attention differential, 3) Möbius band in transfer attack, 4) low sensitivity to attack steps. The code is available at https://github.com/hfmei/VEAttack-LVLM