graph neural networks · AI news

The method shows how graph neural networks can extract clinically meaningful subtypes from population-scale single-cell data, turning donor tissue into a map of resilience and psychiatric comorbidity in AD. Researchers in computational biology and drug discovery can reuse the approach on other diseases where donor cohorts exist. Teams building biomarker pipelines should note the link between cell-level patterns and depression onset in AD patients.

#graph neural networks #Nature Medicine

Nature Medicine@Daifeng WangSource

Papers Builders09:22

Chinese contentEMERGE:首个SE(3)等变图扩散模型实现分辨率无关点云生成

一个做 3D 点云生成的新扩散架构 EMERGE，用图神经网络替代常规网格化管线，一次训练就能在任意分辨率零样本出结果，做 3D 生成的话可以看看这篇。

#EMERGE #diffusion models #graph neural networks

arXiv cs.AI@Ilias Mitsouras et al. (4)Source

Papers12:24

Chinese contentG-NAC：让数据在图上像元胞自动机一样自发聚成域的无监督聚类

一种新的无监督聚类思路：数据点像元胞自动机一样自己"长"出类别，73 个聚类任务上追平了 Genie，还能迁移到大图。

#G-NAC #graph neural networks #Genie

arXiv cs.LG@Keith Miller, Tristan CrawfordSource

Papers Builders09:36

Chinese content新方法提升图神经网络性能，突破传统图同构测试限制

这个研究挺有意思，作者提出的新方法能提升图神经网络在图分类任务上的表现，特别是Graph-UTS增强方法效果不错，值得看看。

#Graph Representation Learning #graph neural networks

arXiv cs.LG@Sanyam Sanjay Jain et al. (4)Source

Papers BuildersSelected09:32

Structural Negative Transfer in Federated Graph Neural Networks

This research provides a detailed analysis of a specific challenge in federated learning for graph data, which is relevant for anyone working on distributed AI systems that process network-structured information.

#Federated Graph Neural Networks #federated learning #graph neural networks

arXiv cs.LG@Chethana Prasad Kabgere, Shylaja SSSource

Papers16:35

Chinese content研究基于集合聚合的循环图神经网络

这篇论文对研究图神经网络和形式化验证的工程师来说很有价值，因为它提供了一种新的方法来构建和解释循环图神经网络。

#Recurrent GNNs #graph neural networks #formal verification

arXiv cs.AI@Blai BonetSource

Papers BuildersSelected73°11:15

GraphFAS for Graph Feature Selection

GraphFAS enables fraud detection systems to work with tabular models and TreeSHAP-based explanations, making it practical for financial institutions in ASEAN.

#GraphFAS #Boruta #graph neural networks

arXiv cs.AI@Yice Luo et al. (10)Source

Papers BuildersSelected10:01

Graph-Agentic RAG for Social Good

The framework provides concrete contracts for building trustworthy AI systems in social-good settings, addressing traceability and oversight concerns.

#Graph-Agentic RAG #RAG #graph neural networks

arXiv cs.AI@Vijay Bommireddy, Raviteja BommireddySource

Papers Builders73°09:51

Chinese content图神经网络中的回声室效应检测

这篇论文揭示了图神经网络中一种新型故障模式，并提出了针对性的解决方案CASP，对GNN研究者和应用开发者很有价值。

#graph neural networks #GNNs

arXiv cs.LG@Asela Hevapathige et al. (4)Source

Papers BuildersSelected73°09:32

GLASS framework for graph-level anomaly detection

GLASS enables zero-shot anomaly detection and few-shot adaptation without target-domain training data, making it valuable for real-world applications with limited labeled data.

#GLASS #graph neural networks #anomaly detection

arXiv cs.LG@Xudong Wang et al. (4)Source

PapersSelected10:07

Edge-Girth for Graph Neural Networks

EGAGNN outperforms cycle-counting methods using fewer parameters, offering a more efficient approach for graph regression tasks.

#graph neural networks #1-WL #ZINC-12k

arXiv cs.LG@Lilian Marey, Charlotte LaclauSource

Papers10:49

Chinese contentMolLedger：基于化学ADME属性的可解释图神经网络

MolLedger用原子级分数解释药物分子预测，比传统方法更符合化学特性，药物研发人员值得关注。

#MolLedger #ADME #graph neural networks #drug discovery

arXiv cs.LG@Christina X. JiSource

PapersSelected73°00:37

Inductive Correlation Clustering with GNNs

The method significantly speeds up clustering for new graph data while maintaining accuracy, making it practical for real-world applications.

#Correlation Clustering #Graph Neural Networks

arXiv cs.LG@Francesco Paolo Nerini et al. (4)Source

PapersSelected00:33

Cone Extended Rayleigh Quotients for Directed Graph Learning

The method improves spectral control in directed graphs while preserving predictive information, outperforming symmetrization approaches that lose directional insights.

#graph neural networks #Cora

arXiv cs.LG@Yavdat Sh. Il'yasov, Nur F. ValeevSource

PapersSelected73°19:32

Graph Neural Networks for Friend Recommendation

The framework reduces memory usage by 98% while improving recommendation performance, making it valuable for large-scale social networks in ASEAN.

#graph neural networks

arXiv cs.LG@Maksim Utushkin et al. (3)Source

Papers10:32

Chinese contentLION：一种用于多模态属性图学习的Clifford神经网络范式

LION基于Clifford代数，在多模态属性图学习上取得了显著成果，值得一看。

#graph neural networks

arXiv cs.LG@Xunkai Li et al. (9)Source