Search Results (29)

Loess deposits in China, covering extensive regions, exhibit distinctive physical and mechanical characteristics, including collapsibility and reduced mechanical strength. These properties contribute to heightened susceptibility to slope-related geological hazards, such as landslides and collapses, in these areas. The widespread distribution and challenging prevention of these geological disasters have emerged as significant impediments to both public safety and economic development in China. Moreover, geological disaster data originates from diverse sources and exists in substantial fragmented, decentralized, and unstructured formats, including textual records and graphical representations. These datasets exhibit complex structures and heterogeneous formats yet suffer from inadequate organization and storage due to the absence of unified descriptive standards. The lack of systematic categorization and standardized representation significantly hinders effective data integration and knowledge extraction across different sources. To address these challenges, this study proposes a novel loess slope stability assessment method employing a dendrogram neural network (GNN-TreeNet) integrated with knowledge graph technology. The methodology progresses through three phases: (1) construction of a multi-domain knowledge graph integrating a large number of loess slopes with historical disaster records, instability factor relationships, and empirical parameter correlations; (2) generation of expressive node embeddings capturing inherent connections via graph neural networks; (3) development and training of the GNN-TreeNet architecture that leverages the graph’s enhanced representation capacity for stability evaluation. This structured framework enables cross-disciplinary data synthesis and interpretable slope stability analysis through a systematic integration of geological, geographical, and empirical knowledge components. Full article

Background: The prognosis management of thyroid cancer remains a significant challenge. This study highlights the critical role of T cells in the tumor microenvironment and aims to improve prognostic precision by integrating bulk RNA-seq and single-cell RNA-seq (scRNA-seq) data, providing a more comprehensive view of tumor biology at the single-cell level. Method: 15 thyroid cancer scRNA-seq samples were analyzed from GEO and 489 patients from TCGA. A multi-level attention graph neural network (MLA-GNN) model was applied to integrate T-cell-related differentially expressed genes (DEGs) for predicting disease-free survival (DFS). Patients were divided into training and validation cohorts in an 8:2 ratio. Result: We systematically characterized the immune microenvironment of metastatic thyroid cancer by using single-cell transcriptomics and identified the important role of T-cell subtypes in the development of thyroid cancer. T-cell-based DEGS between tumor tissues and normal tissues were also identified. Subsequently, T-cell-based risk signatures were selected for establishing a risk model using MLA-GNN. Finally, our MLA-GNN-based model demonstrated an excellent ability to predict the DFS of thyroid cancer patients (1-year AUC: 0.965, 3-years AUC: 0.979, and 5-years AUC: 0.949 in training groups, and 1-year AUC: 0.879, 3-years AUC: 0.804, and 5-years AUC: 0.804 in validation groups). Conclusions: Risk features based on T-cell genes have demonstrated the effectiveness in predicting the prognosis of thyroid cancer. By conducting a comprehensive characterization of T-cell features, we aim to enhance our understanding of the tumor’s response to immunotherapy and uncover new strategies for the treatment of cancer. Full article

Identifying correspondences of road segments in different road networks, namely road-network matching, is an essential task for road network-centric data processing such as data integration of road networks and data quality assessment of crowd-sourced road networks. Traditional road-network matching usually relies on feature engineering and parameter selection of the geometry and topology of road networks for similarity measurement, resulting in poor performance when dealing with dense and irregular road network structures. Recent development of graph neural networks (GNNs) has demonstrated unsupervised modeling power on road network data, which learn the embedded vector representation of road networks through spatial feature induction and topology-based neighbor aggregation. However, weighting spatial information on the node feature alone fails to give full play to the expressive power of GNNs. To this end, this paper proposes a Spatial Pattern-aware Graph EMbedding learning method for road-network matching, named SP-GEM, which explores the idea of spatially-explicit modeling by identifying spatial patterns in neighbor aggregation. Firstly, a road graph is constructed from the road network data, and geometric, topological features are extracted as node features of the road graph. Then, four spatial patterns, including grid, high branching degree, irregular grid, and circuitous, are modelled in a sector-based road neighborhood for road embedding. Finally, the similarity of road embedding is used to find data correspondences between road networks. We conduct an algorithmic accuracy test to verify the effectiveness of SP-GEM on OSM and Tele Atlas data. The algorithmic accuracy experiments show that SP-GEM improves the matching accuracy and recall by at least 6.7% and 10.2% among the baselines, with high matching success rate (>70%), and improves the matching accuracy and recall by at least 17.7% and 17.0%, compared to the baseline GNNs, without spatially-explicit modeling. Further embedding analysis also verifies the effectiveness of the induction of spatial patterns. This study not only provides an effective and practical algorithm for road-network matching, but also serves as a test bed in exploring the role of spatially-explicit modeling in GNN-based road network modeling. The experimental performances of SP-GEM illuminate the path to develop GeoEmbedding services for geospatial applications. Full article

Gene regulatory networks (GRNs) describe the interactions between transcription factors (TFs) and their target genes, playing a crucial role in understanding gene functions and how cells regulate gene expression under different conditions. Recent advancements in multi-omics technologies have provided new opportunities for more comprehensive GRN inference. Among these data types, gene expression and chromatin accessibility are particularly important, as they are key to distinguishing between direct and indirect regulatory relationships. However, existing methods primarily rely on gene expression data while neglecting biological information such as chromatin accessibility, leading to an increased occurrence of false positives in the inference results. To address the limitations of existing approaches, we propose MultiGNN, a supervised framework based on graph neural networks (GNNs). Unlike conventional GRN inference methods, MultiGNN leverages features extracted from both gene expression and chromatin accessibility data to predict regulatory interactions between genes. Experimental results demonstrate that MultiGNN consistently outperforms other methods across seven datasets. Additionally, ablation studies validate the effectiveness of our multi-omics feature integration strategy, offering a new direction for more accurate GRN inference. Full article

Graphs are an effective data structure for characterizing ubiquitous connections as well as evolving behaviors that emerge in inter-wined systems. Limited by the stereotype of node-to-node connections, learning node representations is often confined in a graph diffusion process where local information has been excessively aggregated, as the random walk of graph neural networks (GNN) explores far-reaching neighborhoods layer-by-layer. In this regard, tremendous efforts have been made to alleviate feature over-smoothing issues such that current backbones can lend themselves to be used in a deep network architecture. However, compared to designing a new GNN, less attention has been paid to underlying topology by graph re-wiring, which mitigates not only flaws of the random walk but also the over-smoothing risk incurred by reducing unnecessary diffusion in deep layers. Inspired by the notion of non-local mean techniques in the area of image processing, we propose a non-local information exchange mechanism by establishing an express connection to the distant node, instead of propagating information along the (possibly very long) original pathway node-after-node. Since the process of seeking express connections throughout a graph can be computationally expensive in real-world applications, we propose a re-wiring framework (coined the express messenger wrapper) to progressively incorporate express links in a non-local manner, which allows us to capture multi-scale features without using a very deep model; our approach is thus free of the over-smoothing challenge. We integrate our express messenger wrapper with existing GNN backbones (either using graph convolution or tokenized transformer) and achieve a new record on the Roman-empire dataset as well as in terms of SOTA performance on both homophilous and heterophilous datasets. Full article

Multimodal sentiment analysis faces a number of challenges, including modality missing, modality heterogeneity gap, incomplete datasets, etc. Previous studies usually adopt schemes like meta-learning or multi-layer structures. Nevertheless, these methods lack interpretability for the interaction between modalities. In this paper, we constructed a new dataset, SM-MSD, for sentiment analysis in social media (SAS) that differs significantly from conventional corpora, comprising 10K instances of diverse data from Twitter, encompassing text, emoticons, emojis, and text embedded in images. This dataset aims to reflect authentic social scenarios and various emotional expressions, and provides a meaningful and challenging evaluation benchmark for multimodal sentiment analysis in specific contexts. Furthermore, we propose a multi-task framework based on heterogeneous graph neural networks (H-GNNs) and contrastive learning. For the first time, heterogeneous graph neural networks are applied to multimodal sentiment analysis tasks. In the case of additional labeling data, it guides the emotion prediction of the missing mode. We conduct extensive experiments on multiple datasets to verify the effectiveness of the proposed scheme. Experimental results demonstrate that our proposed scheme surpasses state-of-the-art methods by 1.7% and 0 in accuracy and 1.54% and 4.9% in F1-score on the MOSI and MOSEI datasets, respectively, and exhibits robustness to modality missing scenarios. Full article

The application of machine learning (ML) for the automatic classification of building elements is a powerful technique for ensuring information integrity in building information models (BIMs). Previous work has demonstrated the favorable performance of such models on classification tasks using geometric information. This research explores the hypothesis that incorporating contextual information into the ML models can improve classification accuracy. To test this, we created a graph data structure where each building element is represented as a node assigned with basic geometric information. The connections between the graph nodes (edges) represent the immediate neighbors of that node, capturing the contextual information expressed in the BIM model. We devised a process for extracting graphs from BIM files and used it to construct a graph dataset of over 42,000 building elements and used the data to train several types of ML models. We compared the classification results of models that rely only on geometry, to graph neural networks (GNNs) that leverage contextual information. This work demonstrates that graph-based models for building element classification generally outperform classic ML models. Furthermore, dividing the graphs that represent complete buildings into smaller subgraphs further improves classification accuracy. These results underscore the potential of leveraging contextual information via graphs for advancing ML capabilities in the BIM environment. Full article

The image texture features obtained from ¹⁸F-fluorodeoxyglucose positron emission tomography/computed tomography (¹⁸F-FDG PET/CT) images of non-small cell lung cancer (NSCLC) have revealed tumor heterogeneity. A combination of genomic data and radiomics may improve the prediction of tumor prognosis. This study aimed to predict NSCLC metastasis using a graph neural network (GNN) obtained by combining a protein–protein interaction (PPI) network based on gene expression data and image texture features. ¹⁸F-FDG PET/CT images and RNA sequencing data of 93 patients with NSCLC were acquired from The Cancer Imaging Archive. Image texture features were extracted from ¹⁸F-FDG PET/CT images and area under the curve receiver operating characteristic curve (AUC) of each image feature was calculated. Weighted gene co-expression network analysis (WGCNA) was used to construct gene modules, followed by functional enrichment analysis and identification of differentially expressed genes. The PPI of each gene module and genes belonging to metastasis-related processes were converted via a graph attention network. Images and genomic features were concatenated. The GNN model using PPI modules from WGCNA and metastasis-related functions combined with image texture features was evaluated quantitatively. Fifty-five image texture features were extracted from ¹⁸F-FDG PET/CT, and radiomic features were selected based on AUC (n = 10). Eighty-six gene modules were clustered by WGCNA. Genes (n = 19) enriched in the metastasis-related pathways were filtered using DEG analysis. The accuracy of the PPI network, derived from WGCNA modules and metastasis-related genes, improved from 0.4795 to 0.5830 (p < 2.75 × 10⁻¹²). Integrating PPI of four metastasis-related genes with ¹⁸F-FDG PET/CT image features in a GNN model elevated its accuracy over a without image feature model to 0.8545 (95% CI = 0.8401–0.8689, p-value < 0.02). This model demonstrated significant enhancement compared to the model using PPI and ¹⁸F-FDG PET/CT derived from WGCNA (p-value < 0.02), underscoring the critical role of metastasis-related genes in prediction model. The enhanced predictive capability of the lymph node metastasis prediction GNN model for NSCLC, achieved through the integration of comprehensive image features with genomic data, demonstrates promise for clinical implementation. Full article

A growing number of studies have shown that aberrant microRNA (miRNA) expression is closely associated with the evolution and development of various complex human diseases. These key biomarkers’ identification and observation are significant for gaining a deeper understanding of disease pathogenesis and therapeutic mechanisms. Consequently, pinpointing potential miRNA–disease associations (MDA) has become a prominent bioinformatics subject, encouraging several new computational methods given the advances in graph neural networks (GNN). Nevertheless, these existing methods commonly fail to exploit the network nodes’ global feature information, leaving the generation of high-quality embedding representations using graph properties as a critical unsolved issue. Addressing these challenges, we introduce the DAEMDA, a computational method designed to optimize the current models’ efficacy. First, we construct similarity and heterogeneous networks involving miRNAs and diseases, relying on experimentally corroborated miRNA–disease association data and analogous information. Then, a newly-fashioned parallel dual-channel feature encoder, designed to better comprehend the global information within the heterogeneous network and generate varying embedding representations, follows this. Ultimately, employing a neural network classifier, we merge the dual-channel embedding representations and undertake association predictions between miRNA and disease nodes. The experimental results of five-fold cross-validation and case studies of major diseases based on the HMDD v3.2 database show that this method can generate high-quality embedded representations and effectively improve the accuracy of MDA prediction. Full article

Identifying similar network structures is key to capturing graph isomorphisms and learning representations that exploit structural information encoded in graph data. This work shows that ego networks can produce a structural encoding scheme for arbitrary graphs with greater expressivity than the Weisfeiler–Lehman (1-WL) test. We introduce $\mathrm{I}\mathrm{G}\mathrm{E}\mathrm{L}$, a preprocessing step to produce features that augment node representations by encoding ego networks into sparse vectors that enrich message passing (MP) graph neural networks (GNNs) beyond 1-WL expressivity. We formally describe the relation between $\mathrm{I}\mathrm{G}\mathrm{E}\mathrm{L}$ and 1-WL, and characterize its expressive power and limitations. Experiments show that $\mathrm{I}\mathrm{G}\mathrm{E}\mathrm{L}$ matches the empirical expressivity of state-of-the-art methods on isomorphism detection while improving performance on nine GNN architectures and six graph machine learning tasks. Full article

Implicit sentiment identification is a significant classical task in text analysis. Graph neural networks (GNNs) have recently been successful in implicit sentiment identification, but the current approaches still suffer from two problems. On the one hand, there is a lack of structural information carried by the single-view graph structure of implicit sentiment texts to accurately capture obscure sentiment expressions. On the other hand, the predefined fixed graph structure may contain some noisy edges that cannot represent semantic information using an accurate topology, which can seriously impair the performance of implicit sentiment analysis. To address these problems, we introduce a knowledge-fusion-based iterative graph structure learning framework (KIG). Specifically, for the first problem, KIG constructs graph structures based on three views, namely, co-occurrence statistics, cosine similarity, and syntactic dependency trees through prior knowledge, which provides rich multi-source information for implicit sentiment analysis and facilitates the capture of implicit obscure sentiment expressions. To address the second problem, KIG innovatively iterates the three original graph structures and searches for their implicit graph structures to better fit the data themselves to optimize the downstream implicit sentiment analysis task. We compared our method with the mainstream implicit sentiment identification methods on two publicly available datasets, and ours outperformed both benchmark models. The accuracy, recall, and F1 values of KIG on the Pun of the Day dataset reached 89.2%, 93.7%, and 91.1%, respectively. Extensive experimental results demonstrate the superiority of our proposed method for the implicit sentiment identification task. Full article

Real-world objects are usually defined in terms of their own relationships or connections. A graph (or network) naturally expresses this model though nodes and edges. In biology, depending on what the nodes and edges represent, we may classify several types of networks, gene–disease associations (GDAs) included. In this paper, we presented a solution based on a graph neural network (GNN) for the identification of candidate GDAs. We trained our model with an initial set of well-known and curated inter- and intra-relationships between genes and diseases. It was based on graph convolutions, making use of multiple convolutional layers and a point-wise non-linearity function following each layer. The embeddings were computed for the input network built on a set of GDAs to map each node into a vector of real numbers in a multidimensional space. Results showed an AUC of 95% for training, validation, and testing, that in the real case translated into a positive response for 93% of the Top-15 (highest dot product) candidate GDAs identified by our solution. The experimentation was conducted on the DisGeNET dataset, while the DiseaseGene Association Miner (DG-AssocMiner) dataset by Stanford’s BioSNAP was also processed for performance evaluation only. Full article

Graph neural networks (GNNs) have achieved remarkable success in structured prediction, owing to the GNNs’ powerful ability in learning expressive graph representations. However, most of these works learn graph representations based on a static graph constructed by an existing parser, suffering from two drawbacks: (1) the static graph might be error-prone, and the errors introduced in the static graph cannot be corrected and might accumulate in later stages, and (2) the graph construction stage and graph representation learning stage are disjoined, which negatively affects the model’s running speed. In this paper, we propose a joint-learning-based dynamic graph learning framework and apply it to two typical structured prediction tasks: syntactic dependency parsing, which aims to predict a labeled tree, and semantic dependency parsing, which aims to predict a labeled graph, for jointly learning the graph structure and graph representations. Experiments are conducted on four datasets: the Universal Dependencies 2.2, the Chinese Treebank 5.1, the English Penn Treebank 3.0 in 13 languages for syntactic dependency parsing, and the SemEval-2015 Task 18 dataset in three languages for semantic dependency parsing. The experimental results show that our best-performing model achieves a new state-of-the-art performance on most language sets of syntactic dependency and semantic dependency parsing. In addition, our model also has an advantage in running speed over the static graph-based learning model. The outstanding performance demonstrates the effectiveness of the proposed framework in structured prediction. Full article

Graph-embedding learning is the foundation of complex information network analysis, aiming to represent nodes in a graph network as low-dimensional dense real-valued vectors for the application in practical analysis tasks. In recent years, the study of graph network representation learning has received increasing attention from researchers, and, among them, graph neural networks (GNNs) based on deep learning are playing an increasingly important role in this field. However, the fact that higher-order neighborhood information cannot be used effectively is a problem of most existing graph neural networks. Moreover, it tends to ignore the influence of latent representation and structural properties on graph embedding. In hopes of solving these issues, we introduce centrality encoding to learn the node properties, add an attention mechanism consideration to better distinguish the significance of neighboring nodes, and introduce random walk regularization to make sample neighbors that consistently satisfy predetermined criteria. This allows us to learn a representation of a potential node. We tested the performance of our model on node-clustering and link prediction tasks using three widely recognized benchmark datasets. The outcomes of our experiments demonstrate that our model significantly surpasses the baseline method in both tasks, indicating that the graph embedding it generates is highly expressive. Full article

A method for training and white boxing of deep learning (DL) binary decision trees (BDT), random forest (RF) as well as mind maps (MM) based on graph neural networks (GNN) is proposed. By representing DL, BDT, RF, and MM as graphs, these can be trained by GNN. These learning architectures can be optimized through the proposed method. The proposed method allows representation of the architectures with matrices because the learning architecture can be expressed with graphs. These matrices and graphs are visible, which makes the learning processes visible, and therefore, more accountable. Some examples are shown here to highlight the usefulness of the proposed method, in particular, for learning processes and for ensuring the accountability of DL together with improvement in network architecture. Full article