Tag: graph neural networks

A new multimodal AI system called HMTF-Net combines text, conversation context, user interaction graphs, images and metadata to detect cyberbullying, ...

A computational framework identified state-backed influence operation accounts with an average F1-score of 97.5 percent and predicted future driver-user interactions ...

A new special issue of Data Science and Engineering collects eleven peer-reviewed papers on data spaces, data governance, and data-driven ...

A new dynamic graph neural framework called GAMMA-Traffic updates road-network connections in real time using cross-correlation analysis, reporting up to ...

A new graph neural network framework called EGNN expands rare attack samples and uses threshold-controlled pseudo-labeling to improve intrusion detection ...

A study of the CICIDS2017 DDoS capture shows node-degree graph features fail under NAT identity collapse while edge-weight concentration separates ...

A new recommender model called COMPASS learns a user-specific allocation over spectral bands instead of a fixed truncation rank, improving ...

A controlled study of 3,600 runs finds normalization schemes help mainly at shallow depths while graph homophily predicts how accuracy ...

Researchers combined greedy kernel interpolation with trainable deep kernels and found the resulting models matched or beat neural networks in ...

A new graph-based framework groups similar learners and augments sparse data to improve cognitive diagnosis predictions across three public datasets.

Researchers have built a Spark GraphFrames framework that enumerates millions of transaction cycles and filters them with timestamp and amount ...

A new contrastive learning framework called VMCL improves knowledge graph embedding for unseen entities and shows smooth, predictable gains as ...

A controlled comparison of ten graph neural network architectures finds that PNA, GAT, GCN, and EvolveGCN produce embedding spaces associated ...

A new recommender framework replaces single user and item embeddings with shared multi-aspect pools, improving accuracy across six datasets.

A new open-access study in Discover Computing presents CMCBI, a recommendation framework that aligns corporate financial signals with consumer behavior ...

A new study reports that a transparent, feature-based framework can detect rumors on social media with accuracy close to that ...

Researchers have introduced HOFA, a contrastive learning framework that mines triangular motifs in social networks and aligns embedding features to ...

Researchers have introduced cMSEDN, a deterministic method that turns multiscale EDN sequence representations into sequence-specific node embeddings for De Bruijn ...

A new open-access study reports that an attention graph neural network tuned by adaptive optimization reached 98 percent accuracy in ...

A Physics-Regularized Graph Neural Network learns collective motion dynamics from trajectory data, achieving lower closed-loop position errors than five baselines ...