Break the Visual Perception: Adversarial Attacks Targeting Encoded Visual Tokens of Large Vision-Language Models
Wang, Yubo, Liu, Chaohu, Qu, Yanqiu, Cao, Haoyu, Jiang, Deqiang, Xu, Linli
–arXiv.org Artificial Intelligence
Large vision-language models (LVLMs) integrate visual information into large language models, showcasing remarkable multi-modal conversational capabilities. However, the visual modules introduces new challenges in terms of robustness for LVLMs, as attackers can craft adversarial images that are visually clean but may mislead the model to generate incorrect answers. In general, LVLMs rely on vision encoders to transform images into visual tokens, which are crucial for the language models to perceive image contents effectively. Therefore, we are curious about one question: Can LVLMs still generate correct responses when the encoded visual tokens are attacked and disrupting the visual information? To this end, we propose a non-targeted attack method referred to as VT-Attack (Visual Tokens Attack), which constructs adversarial examples from multiple perspectives, with the goal of comprehensively disrupting feature representations and inherent relationships as well as the semantic properties of visual tokens output by image encoders. Using only access to the image encoder in the proposed attack, the generated adversarial examples exhibit transferability across diverse LVLMs utilizing the same image encoder and generality across different tasks. Extensive experiments validate the superior attack performance of the VT-Attack over baseline methods, demonstrating its effectiveness in attacking LVLMs with image encoders, which in turn can provide guidance on the robustness of LVLMs, particularly in terms of the stability of the visual feature space.
arXiv.org Artificial Intelligence
Oct-9-2024
- Country:
- Oceania > Australia
- North America
- United States
- Texas > Dallas County
- Dallas (0.04)
- New York > New York County
- New York City (0.04)
- Louisiana > Orleans Parish
- New Orleans (0.04)
- Hawaii > Honolulu County
- Honolulu (0.04)
- California
- Santa Clara County > San Jose (0.04)
- San Francisco County > San Francisco (0.04)
- San Diego County > San Diego (0.04)
- Los Angeles County > Long Beach (0.04)
- Texas > Dallas County
- Canada
- Quebec > Montreal (0.04)
- British Columbia > Metro Vancouver Regional District
- Vancouver (0.14)
- Alberta > Census Division No. 15
- Improvement District No. 9 > Banff (0.04)
- United States
- Europe
- Austria (0.04)
- United Kingdom > England
- Greater London > London (0.04)
- Portugal > Lisbon
- Lisbon (0.04)
- France > Île-de-France
- Asia
- Middle East > UAE
- Abu Dhabi Emirate > Abu Dhabi (0.04)
- China > Anhui Province
- Hefei (0.05)
- Middle East > UAE
- Africa > Central African Republic
- Ombella-M'Poko > Bimbo (0.04)
- Genre:
- Research Report > New Finding (0.46)
- Industry:
- Information Technology > Security & Privacy (0.66)
- Transportation (0.46)
- Government > Military (0.42)
- Technology:
- Information Technology > Artificial Intelligence
- Vision (1.00)
- Natural Language
- Large Language Model (1.00)
- Chatbot (0.71)
- Machine Learning
- Neural Networks > Deep Learning (1.00)
- Statistical Learning (0.93)
- Information Technology > Artificial Intelligence