Revisiting CroPA: A Reproducibility Study and Enhancements for Cross-Prompt Adversarial Transferability in Vision-Language Models
重新审视CroPA:面向视觉-语言模型跨提示对抗转移性的可重复性研究与改进
机构 * Mehta Family School of Data Science and Artificial Intelligence(梅hta家族数据科学与人工智能学院) ; Indian Institute of Technology, Roorkee(印度理工学院罗奥克学院) ; Department of Civil Engineering(土木工程系) ; Department of Electronics and Communication(电子与通信系)
专题命中 视觉问答 :vision-language model(title,abstract);LLaVA(abstract,abstract_cn);visual question answering(abstract);分类 cs.CV
AI总结 本文重新审视CroPA,验证其跨提示转移性,并提出改进方法,包括新的初始化策略、跨图像迁移性研究及针对视觉编码器的损失函数,提升对抗有效性。
Comments Accepted to MLRC 2025
Journal ref Transactions on Machine Learning Research (TMLR), 2025. Available at OpenReview: https://openreview.net/forum?id=5L90cl0xtf