Jul 2026· Journal of King Saud University: Computer and Information Sciences· Vol 38· 0 citations· 48 references
TL;DR
GCNCDNMF, a framework that integrates graph convolutional networks (GCNs) with a constrained deep nonnegative matrix factorization (CDNMF) module, which consistently outperforms state-of-the-art methods in accuracy, normalized mutual information, and adjusted Rand index.
Abstract
Community detection is a key task for revealing functional organization in complex networks. Graph neural networks (GNNs) capture non-linear relationships but often suffer from over-smoothing as layers increase. Deep nonnegative matrix factorization (DNMF) models are interpretable via hierarchical learning but are linear and sensitive to topological noise. We propose GCNCDNMF, a framework that integrates graph convolutional networks (GCNs) with a constrained deep nonnegative matrix factorization (CDNMF) module. The CDNMF component uses a deep autoencoder-like structure with graph regularization to preserve community structures. The hierarchical embeddings from CDNMF are injected into the GCN propagation steps, which mitigates over-smoothing. In return, the GCN’s non-linear reconstructions refine the network topology and feed back to CDNMF, improving noise robustness. Experiments on five real-world benchmark datasets (Cora, Citeseer, Email, Cornell, and Texas) show that GCNCDNMF consistently outperforms state-of-the-art methods in accuracy, normalized mutual information, and adjusted Rand index. The code for this work is publicly available at: https://github.com/LiShunli0719/GCNCDNMF_main.
This work introduces ECHO (Encoding Communities via High-Order Operators), a scalable, self-supervised framework that treats communities as regions of adaptive diffusion on semantic manifolds that recovers communities from topology when node features are weak, while feature isolation is preferable when features are strong.
Graph convolutional networks (GCNs) have shown promising performance in real-world applications involving graph-structured data. To enhance the model’s ability to integrate information from neighbor nodes, GCNs typically expand the receptive field by stacking more layers, which can lead to the problem of over-smoothing. To address this limitation, high-order graph convolutional networks (HGCNs) such as Mixhop have been proposed. These high-order models, however, introduce the risk of overfitting due to the increased number of parameters required for designing different order convolutions, which in turn can degrade classification performance. To overcome these issues, we first provide a theoretical analysis of the bottlenecks in conventional GCNs and establish that an effective graph convolutional model should integrate multi-hop neighborhood information in a unified manner. Inspired by this, we propose a novel broad graph convolutional network (BGCN) paradigm, which completely eliminates hidden layers and instead expands the receptive field through network width. We introduce four specific BGCN architectures and provide a detailed analysis of their computational complexity and parameter efficiency. A number of experiments on citation networks demonstrate that the proposed methods achieve state-of-the-art performance in terms of both classification accuracy and stability. In particular, our BGCN with simple attention mechanism (
$$\text {BGCN}_{\text {SA}}$$
BGCN
SA
) achieves top performance on the Citeseer, Cora, and Pubmed datasets, with accuracies of 73.3%, 83.5%, and 81.1%, respectively. The main advantage of this work lies in offering a robust, efficient, and scalable alternative for graph representation learning, effectively circumventing the inherent limitations of over-smoothing and overfitting in existing deep and high-order GCN models.
Alex Hay-Man Ng, Xun Liu, Fangyuan Lei et al.· Complex & Intelligent Sy...· 0 citations
A diffusion-enhanced inductive link prediction framework that combines Graph Diffusion Convolution (GDC), structural node descriptors, and neighborhood aggregation from GraphSAGE is proposed that achieves higher accuracy than the other models on the benchmark datasets.
A combined HGT-based framework incorporating contrastive representation learning and deep clustering with multi-round training via pseudo-labels is presented, enabling the model to better deal with label scarcity, heterogeneous dependencies, and overlapping semantics in practical, complex, attributed networks.
Hamza Haddad, Hicham Attariuas, A. Younes· Edelweiss Applied Science an...· 0 citations
SimGAT, a structure-aware graph attention model built on SimRank-derived structural embeddings, is proposed, which computes structural similarity in the SimRank2Vec embedding space and injects it as a topological prior into the graph attention mechanism, enabling neighborhood aggregation to be jointly guided by node attributes and global structural relationships.
Chengda Xu, Yinglong Zhang· Journal of King Saud Univers...· 0 citations
Graph autoencoders (GAEs) are widely used for learning representations of dynamic graphs. However, their optimisation objectives typically do not take structural heterogeneity across nodes into account. We propose three distance-based GAE variants that incorporate structural penalties into the reconstruction loss. All variants share a two-layer Graph Convolutional Network encoder and a Euclidean-distance decoder trained with distance-based reconstruction objectives. We extend sparsity-corrected loss with two node-level regularization terms: (i) a hub penalty based on degree centrality, and (ii) a penalty based on Natural Community Local Intrinsic Dimensionality (NC-LID). The paper is motivated by prior evidence linking high NC-LID to reduced embedding quality. The proposed methods are designed to emphasize reconstruction errors for structurally ambiguous nodes. Experiments on multiple dynamic graph data sets show that incorporating NC-LID-based regularization consistently improves reconstruction performance over the baseline without structural regularization and the method using hub-aware regularization. These findings highlight NC-LID as a useful structural signal for enhancing distance-based graph autoencoders in dynamic settings.
Aleksandar Tomčić, Milos Savic, Milos Radovanovic· 0 citations