English(EN) Graph Representation Learning Augmented Model Manipulation on Federated Fine-Tuning of LLMs

新的AugMP策略针对大语言模型的联邦微调

作者 PulseAugur 编辑部 · [1 个来源] · 2026-05-08 16:24

研究人员开发了一种名为增强模型操纵（AugMP）的新策略，用于攻击大语言模型（LLMs）的联邦微调（FFT）。该方法利用图表示学习识别合法LLM更新中的相关性，然后指导创建恶意更新。一种迭代算法优化这些恶意更新，以嵌入对抗性目标，同时使其看起来与良性更新相似，从而难以检测。实验表明，AugMP可以显著降低全局LLM准确性和本地代理性能，同时规避标准防御机制。 AI

影响引入了一种新的攻击向量，可能危及通过联邦学习训练的LLM的完整性。

排序理由详细介绍LLM新型攻击方法的学术论文。[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 →

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; 详细介绍LLM新型攻击方法的学术论文。[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: 121 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]

arXiv cs.LG TIER_1 English(EN) · Ozgur B. Akan · 2026-05-08 16:24

基于图表示学习的联邦微调大模型增强模型操纵

Federated fine-tuning (FFT) has emerged as a privacy-preserving paradigm for collaboratively adapting large language models (LLMs). Built upon federated learning, FFT enables distributed agents to jointly refine a shared pretrained LLM by aggregating local LLM updates without sha…

本文如何被排名

报道来源 [1]

基于图表示学习的联邦微调大模型增强模型操纵

相关实体

相关话题