A Comprehensive Survey on Deep Graph Representation Learning

Ju, Wei; Fang, Zheng; Gu, Yiyang; Liu, Zequn; Long, Qingqing; Qiao, Ziyue; Qin, Yifang; Shen, Jianhao; Sun, Fang; Xiao, Zhiping; Yang, Junwei; Yuan, Jingyang; Zhao, Yusheng; Wang, Yifan; Luo, Xiao; Zhang, Ming

doi:10.1016/j.neunet.2024.106207

Cited by 33 publications

(3 citation statements)

References 311 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It incorporates a bias towards selecting the next node during the generation of node sequences. Recently, the rapid development of deep learning has greatly promoted the effective representation of graph data by graph neural networks, leading to significant advancements and improvements in graph-based tasks [32]. Among them, Graph Convolutional Networks (GCNs) [27] utilize the adjacency matrix to capture node relationships, propagating and aggregating features based relationships through multiple layers.…”

Section: Graph Deep Learning Methodsmentioning

confidence: 99%

Integrating Neighborhood Geographic Distribution and Social Structure Influence for Social Media User Geolocation

Zhang

2024

CMES

View full text Add to dashboard Cite

Geolocating social media users aims to discover the real geographical locations of users from their publicly available data, which can support online location-based applications such as disaster alerts and local content recommendations. Social relationship-based methods represent a classical approach for geolocating social media. However, geographically proximate relationships are sparse and challenging to discern within social networks, thereby affecting the accuracy of user geolocation. To address this challenge, we propose user geolocation methods that integrate neighborhood geographical distribution and social structure influence (NGSI) to improve geolocation accuracy. Firstly, we propose a method for evaluating the homophily of locations based on the k-order neighborhood geographic distribution (k-NGD) similarity among users. There are notable differences in the distribution of k-NGD similarity between location-proximate and non-location-proximate users. Exploiting this distinction, we filter out non-location-proximate social relationships to enhance location homophily in the social network. To better utilize the location-proximate relationships in social networks, we propose a graph neural network algorithm based on the social structure influence. The algorithm enables us to perform a weighted aggregation of the information of users' multi-hop neighborhood, thereby mitigating the over-smoothing problem of user features and improving user geolocation performance. Experimental results on real social media dataset demonstrate that the neighborhood geographical distribution similarity metric can effectively filter out non-location-proximate social relationships. Moreover, compared with 7 existing social relationship-based user positioning methods, our proposed method can achieve multi-granularity user geolocation and improve the accuracy by 4.84% to 13.28%.

show abstract

Section: Graph Deep Learning Methodsmentioning

confidence: 99%

Integrating Neighborhood Geographic Distribution and Social Structure Influence for Social Media User Geolocation

Zhang

2024

CMES

View full text Add to dashboard Cite

show abstract

“…GNN methods and Open-Government Data (OGD) have been studied for predicting TF incidents [17]. On traffic information in real time OGD, the Temporal Graph Convolutional Network (TGCN) and Diffusion Convolutional Recurrent Neural Network (DCRNN) prototypes superior to the Historical Average and Autoregressive Integrated Moving Average (HAAIMA).…”

Section: Literature Reviewmentioning

confidence: 99%

Enhancing Urban Traffic Management Through Hybrid Convolutional and Graph Neural Network Integration

Mohsin,

Mettu,

Madhuri

et al. 2024

JMC

View full text Add to dashboard Cite

Traffic congestion has made city planning and citizen well-being difficult due to fast city growth and the increasing number of vehicles. Traditional traffic management fails to solve urban transportation's ever-changing issues. Traffic prediction and control systems are vital for enhancing Traffic Flow (TF) and minimizing congestion. Smart cities need advanced prediction models to regulate urban TF as traffic management becomes more complex. This paper introduces a hybrid Convolutional Neural Networks (CNN) and Graph Neural Networks (GNN) model for better real-time traffic management. The hybrid model combines CNNs' spatial feature extraction with GNNs' structural and relational data processing to analyze and predict traffic conditions. Traffic camera images are pre-processed to extract spatial characteristics. Traffic network graph construction is used for structural research. The model accurately captures traffic topology and space. The proposed method sequentially processes spatial data with CNNs and integrates them with GNNs. The final hybrid model is trained on one year of traffic data from diverse circumstances and events. The hybrid model is compared to CNN, GNN, and traditional Traffic Prediction Models (TPM) like ARIMA and SVM utilizing MAE, RMSE, and MAPE. The hybrid GNN+CNN model outperforms benchmark models with lower MAE, RMSE, and MAPE across several prediction intervals.

show abstract

“…Additionally, molecular fingerprints, substructure fingerprints, and 2D compound images generated by the RDKit package were utilized as input features 42 , 43 . These features were then used to train both traditional machine learning algorithms such as support vector machines (SVMs) 44 , 45 , k-nearest neighbors (kNNs) 46 , 47 , random forests 48 , 49 , and naive Bayes classifiers 50 – 52 , as well as deep learning methods including dense neural networks (DNNs) 53 , 54 , 1D convolutional neural networks (CNNs), and 2D CNNs 21 , 38 , 55 .…”

Section: Introductionmentioning

confidence: 99%

Predicting blood–brain barrier permeability of molecules with a large language model and machine learning

Huang,

Yang,

Liao

et al. 2024

Sci Rep

View full text Add to dashboard Cite

Predicting the blood–brain barrier (BBB) permeability of small-molecule compounds using a novel artificial intelligence platform is necessary for drug discovery. Machine learning and a large language model on artificial intelligence (AI) tools improve the accuracy and shorten the time for new drug development. The primary goal of this research is to develop artificial intelligence (AI) computing models and novel deep learning architectures capable of predicting whether molecules can permeate the human blood–brain barrier (BBB). The in silico (computational) and in vitro (experimental) results were validated by the Natural Products Research Laboratories (NPRL) at China Medical University Hospital (CMUH). The transformer-based MegaMolBART was used as the simplified molecular input line entry system (SMILES) encoder with an XGBoost classifier as an in silico method to check if a molecule could cross through the BBB. We used Morgan or Circular fingerprints to apply the Morgan algorithm to a set of atomic invariants as a baseline encoder also with an XGBoost classifier to compare the results. BBB permeability was assessed in vitro using three-dimensional (3D) human BBB spheroids (human brain microvascular endothelial cells, brain vascular pericytes, and astrocytes). Using multiple BBB databases, the results of the final in silico transformer and XGBoost model achieved an area under the receiver operating characteristic curve of 0.88 on the held-out test dataset. Temozolomide (TMZ) and 21 randomly selected BBB permeable compounds (Pred scores = 1, indicating BBB-permeable) from the NPRL penetrated human BBB spheroid cells. No evidence suggests that ferulic acid or five BBB-impermeable compounds (Pred scores < 1.29423E−05, which designate compounds that pass through the human BBB) can pass through the spheroid cells of the BBB. Our validation of in vitro experiments indicated that the in silico prediction of small-molecule permeation in the BBB model is accurate. Transformer-based models like MegaMolBART, leveraging the SMILES representations of molecules, show great promise for applications in new drug discovery. These models have the potential to accelerate the development of novel targeted treatments for disorders of the central nervous system.

show abstract

A Comprehensive Survey on Deep Graph Representation Learning

Cited by 33 publications

References 311 publications

Integrating Neighborhood Geographic Distribution and Social Structure Influence for Social Media User Geolocation

Integrating Neighborhood Geographic Distribution and Social Structure Influence for Social Media User Geolocation

Enhancing Urban Traffic Management Through Hybrid Convolutional and Graph Neural Network Integration

Predicting blood–brain barrier permeability of molecules with a large language model and machine learning

Contact Info

Product

Resources

About