研究人员发现了一个跨模态编码器（如CLIP）的漏洞，该编码器将文本和图像映射到共享的嵌入空间。他们发现，单个“中心文本”可以与许多不相关的图像生成高相似度分数，从而破坏图像字幕和检索等任务的评估指标。这一发现凸显了高维数据中中心性问题带来的实际安全威胁。 AI
影响 揭示了多模态AI系统遭受对抗性攻击的潜在可能性，影响了评估的可靠性。
排序理由 学术论文，详细介绍了一种识别跨模态编码器漏洞的新方法。
AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →
研究人员发现单个中心文本利用了CLIP跨模态编码器的漏洞
报道来源 [3]
-
One Single Hub Text Breaks CLIP: Identifying Vulnerabilities in Cross-Modal Encoders via Hubness
arXiv:2604.27674v1 Announce Type: cross Abstract: The hubness problem, in which hub embeddings are close to many unrelated examples, occurs often in high-dimensional embedding spaces and may pose a practical threat for purposes such as information retrieval and automatic evaluati…
-
One Single Hub Text Breaks CLIP: Identifying Vulnerabilities in Cross-Modal Encoders via Hubness
The hubness problem, in which hub embeddings are close to many unrelated examples, occurs often in high-dimensional embedding spaces and may pose a practical threat for purposes such as information retrieval and automatic evaluation metrics. In particular, since cross-modal simil…
-
One Single Hub Text Breaks CLIP: Identifying Vulnerabilities in Cross-Modal Encoders via Hubness
The hubness problem, in which hub embeddings are close to many unrelated examples, occurs often in high-dimensional embedding spaces and may pose a practical threat for purposes such as information retrieval and automatic evaluation metrics. In particular, since cross-modal simil…