MS student, Mohamed bin Zayed University of Artificial Intelligence
1 paper at NeurIPS 2025
FOA-Attack enhances targeted adversarial transferability to closed-source MLLMs by optimally aligning global and local features.
