None Unextractable Protocol Models: Collaborative Training and Inference without Weight Materialization

新的UPM支持在不提取权重的情况下进行协作式AI训练

作者 PulseAugur 编辑部 · [2 sources] · 2026-05-22 10:24

研究人员推出了一种名为不可提取协议模型（UPMs）的新框架，用于神经网络的协作训练和推理，其中各个参与者仅处理模型的一部分。该方法通过定期注入时变变换，确保任何单个实体都无法获得完整的模型权重集。UPMs在困惑度方面影响极小，并且在推理和训练过程中仅增加少量的延迟、带宽和内存开销。 AI

影响通过防止模型提取，实现安全的协作式AI开发，可能促进社区驱动的训练计划。

排序理由详细介绍AI模型训练和推理新方法的学术论文。

AI 生成摘要 · Google Gemini · 来自 2 个来源。我们如何撰写摘要 →

报道来源 [2]

arXiv cs.LG TIER_1 · Alexander Long, Chamin Hewa Koneputugodage, Thalaiyasingam Ajanthan, Yan Zuo, Gil Avraham, Violetta Shevchenko, Hadi Mohaghegh Dolatabadi, Sameera Ramasinghe · 2026-05-25 04:00

Unextractable Protocol Models: Collaborative Training and Inference without Weight Materialization

arXiv:2605.23464v1 Announce Type: new Abstract: We consider a decentralized setup in which the participants collaboratively train and serve a large neural network, and where each participant only processes a subset of the model. In this setup, we explore the possibility of unmate…
arXiv cs.LG TIER_1 · Sameera Ramasinghe · 2026-05-22 10:24

Unextractable Protocol Models: Collaborative Training and Inference without Weight Materialization

We consider a decentralized setup in which the participants collaboratively train and serve a large neural network, and where each participant only processes a subset of the model. In this setup, we explore the possibility of unmaterializable weights, where a full weight set is n…