Vision-Language Agentic Systems (VLAS) connect visual perception to planning and actions, making visual backdoors a system-level threat. Current evaluations on such backdoors focus on clean accurac...
<p><strong>Tác giả:</strong> Jiamin Chang, Salil S. Kanhere, Piotr Koniusz, Jason, Xue</p>
<p>Vision-Language Agentic Systems (VLAS) connect visual perception to planning and actions, making visual backdoors a system-level threat. Current evaluations on such backdoors focus on clean accuracy and attack success rate (ASR), metrics that capture whether a trigger works, but not whether an attack is actually “precise” – i.e. whether it triggers hidden behaviors only when intended. In this work, we formalize the failure of trigger precision as “trigger leakage”: inputs that are visually or...</p>
<p>📄 DOI: <a href="https://doi.org/https://doi.org/10.1145/3846374.3846397" target="_blank" rel="noopener">https://doi.org/10.1145/3846374.3846397</a></p>
<p>📖 <a href="https://doi.org/https://doi.org/10.1145/3846374.3846397" target="_blank" rel="noopener">Xem bài báo đầy đủ</a></p>
Bài viết được biên soạn lại bởi LIFVietnam. Nguồn tham khảo: OpenAlex
