English(EN) Sinkhorn doubly stochastic attention rank decay analysis

研究发现 Sinkhorn 注意力可保持 Transformer 的秩衰减

作者 PulseAugur 编辑部 · [1 个来源] · 2026-10-07 04:00

研究人员分析了 Sinkhorn 算法在 Transformer 架构中应用于双随机注意力的效果，发现与标准的行随机注意力相比，它能更有效地保持秩。该研究包括理论界限和在情感分析及图像分类任务上的实证验证，表明跳跃连接对于缓解秩崩溃至关重要，秩崩溃是指随着网络深度的增加，token 表示变得越来越均匀。研究表明，在使用 Sinkhorn 归一化时，秩会随着深度的增加呈双指数衰减，这与标准 softmax 注意力的发现类似。 AI

影响为注意力机制提供了理论见解，可能指导未来 Transformer 模型开发以提高性能。

排序理由学术论文，详细介绍了对 Transformer 架构中注意力机制的理论和实证分析。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.AI 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: 15 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 学术论文，详细介绍了对 Transformer 架构中注意力机制的理论和实证分析。[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, model release; 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.AI TIER_1 English(EN) · Michela Lapenna, Rita Fioresi, Bahman Gharesifard · 2026-10-07 04:00

Sinkhorn 双随机注意力秩衰减分析

arXiv:2604.07925v2 Announce Type: replace-cross Abstract: The self-attention mechanism is central to the success of Transformer architectures. However, standard row-stochastic attention has been shown to suffer from significant signal degradation across layers. In particular, it …

本文如何被排名

报道来源 [1]

Sinkhorn 双随机注意力秩衰减分析

相关实体

相关话题