PulseAugur
中
实时 15:50:03
English(EN) Poison-then-Hide: Finetuning-Activated Backdoor Attack on Pretrained Vision Encoders

新的“Poison-then-Hide”攻击绕过了微调视觉编码器的防御机制

研究人员开发了一种名为 Poison-then-Hide 的新型后门攻击,该攻击可以破坏预训练的视觉编码器。即使在干净的下游数据上对整个模型进行微调的情况下,这种攻击仍然有效,而之前的攻击在这种情况下常常失败。Poison-then-Hide 的工作原理是在基础编码器中植入一个休眠的后门,优化一个触发器,然后使用有针对性的遗忘技术来隐藏后门。该攻击在六个数据集和三种模型架构上取得了高达 97% 的成功率,突显了模型在这种复杂攻击面前的脆弱性,因为标准防御措施无法完全清除后门。 AI

影响 这项研究突显了视觉编码器微调过程中存在的重大漏洞,可能影响已部署的 AI 系统的安全性。

排序理由 该集群描述了一篇新颖的学术研究论文,详细介绍了一种针对机器学习模型的新型攻击。 [lever_c_demoted from research: ic=1 ai=1.0]

在 Hugging Face Daily Papers 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“Poison-then-Hide”攻击绕过了微调视觉编码器的防御机制

本文如何被排名

Signal score
0 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
该集群描述了一篇新颖的学术研究论文,详细介绍了一种针对机器学习模型的新型攻击。 [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
6 days old
Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.

完整方法见我们的编辑标准。

报道来源 [1]

  1. Hugging Face Daily Papers TIER_1 English(EN) ·

    投毒后隐藏:预训练视觉编码器的微调激活后门攻击

    Backdoor attacks threaten the integrity of machine learning models by allowing attackers to control model behavior through triggers. Models can be compromised during finetuning by backdoor attacks through either poisoned data or adversarial training objectives. Because existing f…