English(EN) Divisive Normalization Shapes Low-Rank Slow Manifolds for Continuous Working Memory

新的RDNN模型使用分裂归一化实现连续工作记忆

作者 PulseAugur 编辑部 · [1 个来源] · 2026-08-25 04:00

研究人员引入了递归分裂归一化网络（RDNN），这是一种受生物分裂归一化启发的创新模型，旨在解决人工神经网络在连续工作记忆方面的局限性。与通常离散化状态空间的GRU和LSTM等传统RNN不同，RDNN可以学习稳健、高保真的慢流形以维持连续变量。该模型机制涉及通过时间反向传播过程中的活动依赖梯度缩放，有效压缩网络的秩并将动力学限制在低维子空间内，从而防止优化问题。这种生物物理约束被证明对于学习连续表示和防止动态输入下的流形破碎至关重要。 AI

影响引入了一种受生物计算启发的创新神经网络架构，有望提高AI系统中连续记忆的稳定性和保真度。

排序理由介绍新模型和分析的学术论文。[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: 1 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 介绍新模型和分析的学术论文。[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, 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: 1 days old; Coverage has settled into its steady-state source set.

完整方法见我们的编辑标准。

报道来源 [1]

arXiv cs.AI TIER_1 English(EN) · Zhaotian Gu, Jie Su, Weiwei Wang, Chang Liu, Tianyi Qian, Dahui Wang · 2026-08-25 04:00

分裂的归一化为连续工作记忆塑造低秩慢流形

arXiv:2608.01947v2 Announce Type: replace-cross Abstract: The ability to robustly maintain and update continuous variables is a hallmark of working memory. While classical continuous attractor networks suffer from severe fine-tuning fragility, standard artificial recurrent neural…

本文如何被排名

报道来源 [1]

分裂的归一化为连续工作记忆塑造低秩慢流形

相关实体

相关话题