Zihao Zhao

AIR achieves unprecedented attack success rates against subject-agnostic face swapping models while maintaining high visual fidelity, without needing a surrogate model.

OrchRM slashes training costs while boosting orchestration accuracy, proving that self-supervised reward modeling can revolutionize multi-agent coordination.

Even with multi-agent adjudication, MLLMs still struggle to reliably differentiate visually confounded diseases in a zero-shot setting, highlighting limitations in current agent-based diagnostic approaches.

By explicitly modeling vascular topology, VasGuideNet achieves anatomically consistent Couinaud liver segmentation, surpassing state-of-the-art methods in Dice score and RVD.