PulseAugur
中
实时 12:40:01
Deutsch(DE) Understanding and Mitigating Token-Pruning-Induced Vulnerabilities in VLMs

新的VLM漏洞:令牌修剪加剧恶意内容

研究人员发现了一种与令牌修剪技术相关的视觉语言模型(VLM)的新漏洞。令牌修剪技术通过移除冗余的视觉令牌来加速模型性能。这种被称为“修剪诱导的恶意放大”(Pruning-Induced Malicious Amplification)的漏洞,在移除良性背景令牌后,会导致模型注意力集中在恶意的前景令牌上,从而无意中放大有毒语义。为了应对这一问题,开发了一种名为“安全感知修剪”(Safety-Aware Pruning, SAP)的新即插即用机制。SAP在推理时识别恶意锚点,恢复良性令牌并重新分配注意力,在不牺牲效率或效用的情况下,显著降低了对抗性攻击的成功率。 AI

影响 识别出与优化技术相关的VLM新类漏洞,可能影响多模态AI系统的安全部署。

排序理由 详细介绍VLM新漏洞和缓解策略的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.LG 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。 我们如何撰写摘要 →

新的VLM漏洞:令牌修剪加剧恶意内容

本文如何被排名

Signal score
8 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
详细介绍VLM新漏洞和缓解策略的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
paper, safety
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Same-day
Cluster formed today. Ranking reflects the current source set at time of score.

完整方法见我们的编辑标准。

报道来源 [1]

  1. arXiv cs.LG TIER_1 Deutsch(DE) · Shuailong Wang, Xinyu Lyu, Shengming Yuan, Jingkuan Song, Heng Tao Shen, Lianli Gao ·

    理解和缓解视觉语言模型(VLMs)中由令牌修剪引起漏洞的方法

    arXiv:2610.09703v1 Announce Type: cross Abstract: Token-Pruning accelerates Vision-Language Models by removing redundant visual tokens, yet its safety implications remain underexplored. In this work, we present the first comprehensive safety evaluation of Token-Pruning mechanisms…