---
格式版本: 2
标题: "scDNM-VAE enables directly inspectable deep clustering of single-cell RNA-seq data through signed dendritic gating"
原文链接: "https://arxiv.org/abs/2608.17228"
发布日期: "2026-08-18"
发布时间校准状态: "found"
发布时间需复核: "否"
发布时间来源: "rule:local:strict_original_body"
发布时间证据: "**\\[v1\\]** Tue, 18 Aug 2026 00:33:42 UTC (8,193 KB)"
发布时间校准原因: "规则确认唯一严格发布时间，来源 local:strict_original_body"
发布时间校准置信度: "high"
发布时间候选数量: 18
发布时间严格候选数量: 6
发布时间原页读取状态: "source template page reused from URL open"
发布时间未找到原因: ""
发布时间校准时间: "2026-08-20T15:55:44+08:00"
发布时间仲裁状态: "skipped"
发布时间仲裁尝试次数: 0
发布时间仲裁耗时毫秒: 0
发现时间: "2026-08-20T15:48:16+08:00"
入库时间: "2026-08-20T07:55:44.871Z"
来源平台: "arXiv 学术论文搜索"
搜索渠道: "source_template"
搜索词: "https://arxiv.org/search/?query=model%20fit&searchtype=all"
匹配关键词:
  - "model fit"
相关厂家:
  []
相关专家:
  []
内容类型: "网页"
抓取工具: "Free Fetch + Defuddle"
清洗工具: "Defuddle Markdown + Defuddle/Readability 正文提取"
原始附件:
  []
AI优质: "否"
AI打分: 2
AI分档: "非优质"
AI质检状态: "不通过"
AI打分理由: "该arXiv论文为单细胞RNA测序聚类方法，属生物信息学领域，与超节点/AI Rack/机柜级AI基础设施完全无关，无任何相关技术或商业信号。"
AI质检模型: "ali-deepseek-v4-flash"
AI质检时间: "2026-08-20T16:01:44+08:00"
AI主题相关性: 0
AI来源权威性: 2
AI新颖性: 0
AI技术细节: 0
AI商业部署信号: 0
AI完整性: 0
AI摘要: "scDNM-VAE 将变分自编码器与树突神经元头结合，通过可学习的带符号突触权重和阈值实现可直接检查的单细胞 RNA-seq 深度聚类。"
AI摘要模型: "ali-deepseek-v4-flash"
AI摘要时间: "2026-09-07T03:22:14.505Z"
采集批次: "2026年8月20日14点19分32秒"
采集批次ID: "20260820-141932-079"
去重键: "https://arxiv.org/abs/2608.17228"
---

## Quantitative Biology > Quantitative Methods

## Title:scDNM-VAE enables directly inspectable deep clustering of single-cell RNA-seq data through signed dendritic gating

[View PDF](https://arxiv.org/pdf/2608.17228) [HTML (experimental)](https://arxiv.org/html/2608.17228v1)

> Abstract:Deep clustering models for single-cell RNA sequencing often assign cells through latent or centroid-based mechanisms that are difficult to inspect. We introduce scDNM-VAE (single-cell Dendritic Neuron Model Variational Autoencoder), a deep clustering framework that combines a variational autoencoder with a dendritic neuron-inspired head. Cluster assignments are governed by learnable signed synaptic weights and thresholds: the weight sign determines the direction of a gate's response to a latent coordinate, its magnitude controls steepness, and the weight-threshold pair determines the transition location. The trained clustering function can therefore be inspected directly without fitting a post-hoc explanation model.  
> We benchmark scDNM-VAE on four datasets spanning immune, cortical, cardiac, and hematopoietic cells against scVI followed by KMeans and an MLP-DEC ablation. scDNM-VAE performs better than scVI on PBMC3k, comparably on the Human Heart Cell Atlas and Paul15, and worse on Zeisel, while producing biologically coherent marker-gene signatures. Ablating each cluster's three highest-magnitude synaptic dimensions causes numerically greater reassignment than random-dimension ablation across all datasets, but the margins are modest and negligible on Zeisel. These results show that signed dendritic gating supports competitive clustering with a parameter-inspectable decision function, while indicating that decision-relevant information is distributed across the latent space.

| Comments: |  |
| --- | --- |
| Subjects: | Quantitative Methods (q-bio.QM); Genomics (q-bio.GN) |
| Cite as: | [arXiv:2608.17228](https://arxiv.org/abs/2608.17228) \[q-bio.QM\] |
|  | (or [arXiv:2608.17228v1](https://arxiv.org/abs/2608.17228v1) \[q-bio.QM\] for this version) |
|  | [https://doi.org/10.48550/arXiv.2608.17228](https://doi.org/10.48550/arXiv.2608.17228) |

## Submission history

From: Melih Agraz \[[view email](https://arxiv.org/show-email/f8dc5e22/2608.17228)\]  
**\[v1\]** Tue, 18 Aug 2026 00:33:42 UTC (8,193 KB)

[Which authors of this paper are endorsers?](https://arxiv.org/auth/show-endorsers/2608.17228) | Disable MathJax ([What is MathJax?](https://info.arxiv.org/help/mathjax.html))