English(EN) Capacity Overflow: A Blind Spot for Backdoor Attacks in Vision MoE

新型后门攻击利用视觉 MoE 容量溢出

作者 PulseAugur 编辑部 · [1 个来源] · 2026-08-27 04:00

研究人员发现了一种针对视觉 Transformer 的混合专家（MoE）架构的新漏洞，称为“容量溢出”。该漏洞源于用于管理专家处理预算的依赖于批次的 token 分派机制。所提出的攻击通过在早期 MoE 层注入后门，并在更深的层注入中和器来利用这一点，而中和器在部署规模的批次大小下会被 token 溢出禁用。实验表明，该攻击在规避现有安全审计的同时，成功率很高。 AI

影响凸显了可扩展视觉 MoE 架构中的一个基本安全风险，可能影响先进计算机视觉模型的部署。

排序理由详细介绍 AI 模型中新型安全漏洞的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.CV 阅读 →

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

本文如何被排名

Signal score: 26 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 详细介绍 AI 模型中新型安全漏洞的学术论文。[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: Breaking (< 6h); Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

完整方法见我们的编辑标准。

报道来源 [1]

arXiv cs.CV TIER_1 English(EN) · Xiaocheng Zou, Tiancheng Zheng, Xiaolin Xu, Ruyi Ding · 2026-08-27 04:00

容量溢出：视觉MoE后门攻击的盲点

arXiv:2608.25371v1 Announce Type: new Abstract: Mixture-of-Experts (MoE) has become a prevalent paradigm for scaling Vision Transformers efficiently. To ensure computational scalability and prevent expert overload, Vision MoE architectures employ a capacity-bounded token dispatch…

本文如何被排名

报道来源 [1]

容量溢出：视觉MoE后门攻击的盲点

相关实体

相关话题